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
- :
- Re: Mean Difference with confidence intervals

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
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Mean Difference with confidence intervals

Sep 9, 2019 5:20 PM
(112 views)

So I'm running an experiment that tests rodents consumption of seeds. The results showed that rodents from the control group consumed 898 seeds on average whereas another group of rodents that received treated seeds consumed only 447 seeds on average. I would like to express these results as the phrase, "rodents from the treatment group consumed 50.2% less seeds than those from the control." {1-(447/898)=.502} Calculating the percent change is easy enough, but I would like to build a confidence interval around that relative percent change. Is there a simple way to calculate this or a shortcut in JMP that might display it?

4 REPLIES 4

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Mean Difference with confidence intervals

I believe from your description that this is a case of a simple two-sample *t*-test. Here is an example, using weight versus sex from the Big Class sample data.

The Pooled t Test outline reports the point estimate and the 95% confidence interval estimate for the difference. Can you convert these values to percentage difference?

Learn it once, use it forever!

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Mean Difference with confidence intervals

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Mean Difference with confidence intervals

Select **Analyze** > **Fit Model**. Enter your **response** as **Y** and your **predictor** with **Add** (Effects) and click **Run**. Click the red triangle at the top left and select **Estimates** > **Multiple Comparisons**, select **Comparisons with Control - Dunnett**, and then click **OK**. Select your **control level** and click **OK**. You will get a table like this:

Can you use the differences here for your purpose? Note that you can right-click on the table and select Make Into Data Table. You can now use column formulas to compute values or visualize the table with Graph Builder.

I don't know how to calculate percentages.

Learn it once, use it forever!

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Mean Difference with confidence intervals

Thank you for your proposed solution, but it only gives the confidence interval of the absolute difference. what I'm searching for is the confidence interval of the relative difference.

The phrase, "Rodents that recieved seeds treated in compound A ate 451 less seeds than those offered control seeds.(CI: 3??, 5??)," is not nearly as informative as the phrase, "Rodents that recieved seeds treated in compound A ate 50.2% less seeds than those offered control seeds.(CI: ??%, ??%),"

I could publish this phrase without a confidence interval, but I feel that it would raise questions from a statistically minded review without the interval.

The phrase, "Rodents that recieved seeds treated in compound A ate 451 less seeds than those offered control seeds.(CI: 3??, 5??)," is not nearly as informative as the phrase, "Rodents that recieved seeds treated in compound A ate 50.2% less seeds than those offered control seeds.(CI: ??%, ??%),"

I could publish this phrase without a confidence interval, but I feel that it would raise questions from a statistically minded review without the interval.