topic Re: Help analyzing Master's project - experimental design in SAS Programming

Help analyzing Master's project - experimental design

Adam1985 — Wed, 10 Oct 2012 17:59:52 GMT

Hi,

I am a Master's student in Natural Resources and Environmental Management at the University of Hawaii at Manoa.

For my Master's project, I examined the effect of irrigation and nutrient management practices on soil CO2 emission under tropical conditions while growing sweet corn.

I believe the project was set up as a split split plot (possibly just a split plot?), and consisted of 3 irrigation rates, 2 organic amendments, 3 levels of each amendment, and a control.

Soil CO2 levels were measured twice a week for the duration of the growing season.

I am having trouble figuring out a way to analyze the data, due to each row only having one control, whereas each amendment had 3 levels. I think that each row should have had 3 controls.

I have attached a file that shows the layout of my design.

If anyone knows how to analyze this experiment, I would be greatly appreciative.

Thanks,

Adam

Re: Help analyzing Master's project - experimental design

SteveDenham — Thu, 11 Oct 2012 14:18:20 GMT

Hi Adam,

Analysis isn't the hard part--getting the proper comparisons will be the key.

I think of this as having two factors--irrigation rate and soil treatment. Soil treatment has seven levels, three for bone meal, three for manure, and one control. These treatments are subplots under the whole plot of irrigation rate. You have three reps.

A possible analysis is then (using SAS/STAT 12.1):

proc mixed;

class rep rate trt;

model co2=rate trt rate*trt/ddfm=kr2;

random intercept/subject=rep;

lsmeans rate trt rate*trt;

lsmestimate <THIS WILL DEPEND ON THE COMPARISONS YOU WISH TO MAKE, AND WHETHER THERE IS AN INTERACTION BETWEEN RATE AND TRT. MORE INFO ON THIS WILL MAKE IT EASIER TO CONSTRUCT THESE PARTICULAR COMPARISONS>;

run;

Hope this gets you started. For more on this kind of analysis, get a copy of SAS for Mixed Models, 2nd ed, by Littell et al., or Milliken's Analysis of Messy Data.

Steve Denham