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
- :
- Help with 'Fit Model' for two-way ANOVA with interaction effects

Topic Options

- Start Article
- 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

Jun 7, 2014 10:16 AM
(4541 views)

The JMP 10 Pro sample data set, 'Snapdragon', is listed as an example for conducting two-way ANOVA. I am having trouble using the 'Fit Model' platform to show results on interaction effects. It works very nicely for displaying the results of the additive model, but when the two effects are crossed and added, the output does not display any results. Any insight into why this problem is occurring would be appreciated. Thanks.

1 ACCEPTED SOLUTION

Accepted Solutions

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

Not enough observations to estimate the interaction effect. Add some replication and you'll be able to run the full factorial model. Try the script below to see the difference.

dt=Open**(** "$SAMPLE_DATA/Snapdragon.jmp" **)**<< **Concatenate****(**

Data Table**(** "Snapdragon.jmp" **)**,

Data Table**(** "Snapdragon.jmp" **)**,

Output Table**(** "ExpandedSnapdragon" **)**

**)**;

dt<<**New Column****(** "NewY",

Numeric,

Continuous,

Format**(** "Best", **12** **)**,

Formula**(** :Y + Random Normal**()** **)**,

Set Selected

**)**;

dt<<**Fit Model****(**

Y**(** :NewY **)**,

Effects**(** :Soil, :Block, :Soil * :Block **)**,

Personality**(** Standard Least Squares **)**,

Emphasis**(** Effect Leverage **)**,

Run**(**

:NewY << **{****Lack of Fit****(** **0** **)**, Plot Actual by Predicted**(** **1** **)**,

Plot Regression**(** **0** **)**, Plot Residual by Predicted**(** **1** **)**,

Plot Effect Leverage**(** **1** **)}**

**)**

**)**;

2 REPLIES 2

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

Not enough observations to estimate the interaction effect. Add some replication and you'll be able to run the full factorial model. Try the script below to see the difference.

dt=Open**(** "$SAMPLE_DATA/Snapdragon.jmp" **)**<< **Concatenate****(**

Data Table**(** "Snapdragon.jmp" **)**,

Data Table**(** "Snapdragon.jmp" **)**,

Output Table**(** "ExpandedSnapdragon" **)**

**)**;

dt<<**New Column****(** "NewY",

Numeric,

Continuous,

Format**(** "Best", **12** **)**,

Formula**(** :Y + Random Normal**()** **)**,

Set Selected

**)**;

dt<<**Fit Model****(**

Y**(** :NewY **)**,

Effects**(** :Soil, :Block, :Soil * :Block **)**,

Personality**(** Standard Least Squares **)**,

Emphasis**(** Effect Leverage **)**,

Run**(**

:NewY << **{****Lack of Fit****(** **0** **)**, Plot Actual by Predicted**(** **1** **)**,

Plot Regression**(** **0** **)**, Plot Residual by Predicted**(** **1** **)**,

Plot Effect Leverage**(** **1** **)}**

**)**

**)**;

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

Re: Help with 'Fit Model' for two-way ANOVA with interaction effects

Thanks MS. Your script was very helpful.

Since there is only one replicate for each soil*block combination, I can see why JMP was unable to estimate an interaction effect.