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
- :
- Missing data w/regression testing & calculating si...

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
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jan 16, 2010 12:18 PM
(298 views)

I am looking to run multiple regression on my data and to calculate significance of predictability; I have my data imported and I've figured out how to run multiple regression, but my questions are:

* How do I set it to handle missing data and 'dont' know' responses?

The data is 5 point numeric rating scale for x and y.

* How do I calculate whether a variable is a statistically signifcant predictor? (I am a stats rookie).

Thank you in advance for any help.

Dan

1 REPLY

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

Jan 18, 2010 8:07 AM
(269 views)

1. Use the grand average of the existing data

2. Use your predicted value for the missing data (this of course assumes you have predicted the results prior to capturing the actual data)

3. Do a least squares regression and calculate the missing data point (run the analysis and save the prediction formula). Note: This method can give misleading results; especially with small data sets.

4. Rerun the missing treatment combination (DOE). Or re-collect the missing sampling points. Be aware of potential blocking effects and changes in Unit Structure.

5. Rerun the entire sampling plan.

What is the response variable? Sounds like a categorical/Likert scale type response?

Use fit model and run standard least squares. Enter your model (X's and Y's). Evaluate the output (ANOVA).

Remember the prediction formula is only as good as the knowledge of the context of population from which the sample was obtained, regardless of the statistical significance.