RobertNiles.com
About Robert
Reporting Help
Finding Data on the Internet
Questions for Candidates
Stats Lessons
Mean
Median
Percent
Per capita
Standard Deviation
Margin of Error
Data Analysis
Sample Sizes
Stats Tests
Help Board
Bookstore


The Cartoon Guide to Statistics

This isn't some dumb-downed whitewash. It's a must-read for a beginner taking his or her first steps toward understanding stats.
More information
-->

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

split plot model problem
Message posted by Josh (via 24.108.13.89) on July 19, 2001 at 9:22 PM (ET)

I'm running an experiment looking at the effects of competition, aboveground (AG) and belowground (BG)herbivory in four species of grassland plants.

The AG and BG treatments are applied factorially (2x) in a complete randomized block design. Within those treatments, I have split the plot and applied competition (C)and species (S) treatments factorially (2x4), with two replicates of each species*competition treatment per wholeplot.

My questions are:

-What should the statistical model for this look like? From what I've read, I know that the error term for testing the significance of the block, AG and BG treatments is the AG*BG*block term. I believe that the error term for testing C, S, and AG or BG by C and/or S effects should be the overall error term.

-Where should the replicate be included? Is this simply another partition which reduces the overall MSE term?

-What about the AG*block and BG*block terms? These are not relevant to the experiment, so are they simply contained in the overall error term, or are they combined with AG*BG*block into an error term for testing wholeplot effects?

Thanks for any help that you can provide.


READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: split plot model problem
Message posted by JG (via 128.8.22.15) on July 20, 2001 at 4:04 AM (ET)

I do not fully understand your problem or what you have done. But, the following thoughts may be helpful. In ANOVA we generally make all the threatments orthogonal so that the sum of squares can be split into appropriate pieces. We generally assume that the higher level interactions are zero, otherwise we need to replicate each cell to have an error term. In the complex design that you are doing you are assuming that the lower level interactions are zero because they are confounded with some of you treatments. Did you split your plots in such a way as to preserve orthogonality ? I hope this helps.


Re: split plot model problem
Message posted by JG (via 128.8.23.228) on July 20, 2001 at 7:04 PM (ET)

You may find the following websit useful - (http://members.aol.com/johnp71/javastat.html) .



Your $5 contribution helps cover part the $500 annual cost of keeping this site online.

Niles Online Sites:RobertNiles.comTheme Park InsiderViolinist.com

RobertNiles.com™, the site, content and services 咀opyright 1996-2002, Robert Niles.
All rights reserved. Questions? Comments? Read my Privacy Policy, or E-mail me!