turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- How do you interpret Welch's Test results?

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Sep 29, 2015 1:20 PM
(6846 views)

Since the tests for Unequal Variances indicates that the variances ARE unequal, how do I interpret the Welch's Test results? My take, based on the JMP Help is that since the Prob > F is greater than 0.05, the means of my two sample groups are NOT significantly different.

Solved! Go to Solution.

1 ACCEPTED SOLUTION

Accepted Solutions

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Sep 30, 2015 8:18 AM
(11557 views)

Solution

jeff.kolton1 wrote:

My take, based on the JMP Help is that since the Prob > F is greater than 0.05, the means of my two sample groups are NOT significantly different.

That is true given that you are testing at the 5% significance level. When using p-values to determine the results of a test the general rule is no significant differences were detected if *p* > *α*; in JMP *α* = 0.05 by default. JMP also shows you significance by color-coding the p-values. If you look at the results of your tests for equal variances, you can see that all of the tests are significant at the 5% level. Note that the Levene, Bartlett, and 2-sided F Test are also significant at the 1% level. My guess, is that the color coding, which I think started with JMP 12, distinguishes between the 5% level (red) and 1% level (orange).

I am not aware of any global way to change the default significance level, but I know I have seen and option to select the significance level for tests in specific platforms.

3 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Sep 30, 2015 8:18 AM
(11558 views)

jeff.kolton1 wrote:

My take, based on the JMP Help is that since the Prob > F is greater than 0.05, the means of my two sample groups are NOT significantly different.

That is true given that you are testing at the 5% significance level. When using p-values to determine the results of a test the general rule is no significant differences were detected if *p* > *α*; in JMP *α* = 0.05 by default. JMP also shows you significance by color-coding the p-values. If you look at the results of your tests for equal variances, you can see that all of the tests are significant at the 5% level. Note that the Levene, Bartlett, and 2-sided F Test are also significant at the 1% level. My guess, is that the color coding, which I think started with JMP 12, distinguishes between the 5% level (red) and 1% level (orange).

I am not aware of any global way to change the default significance level, but I know I have seen and option to select the significance level for tests in specific platforms.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Sep 30, 2015 8:37 AM
(6501 views)

You can control the formatting that JMP uses for the p-values. Go to File > Preferences. Click the Reports category. Click the "Manage Rules" button next to Show Conditional Formatting.

From there choose PValue and edit. You then get a dialog box that allows you to change when JMP flags a low p-value.

Note that by using p-values, you never really have to specify your confidence level. You just always use the generic rule: if p-value is less than alpha, reject the null hypothesis. It is the formatting in JMP that may cause issues.

Dan Obermiller

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Sep 30, 2015 11:45 AM
(6501 views)