greenhouse experiment in a rcbd with two inoculation sites per EU and three repeat measurements

[JA57]

I'm a novice at statistics and SAS and need of some advice in analyzing greenhouse data.  I am testing the efficacy of a single biocontrol on a plant pathogen.  I have one plant per treatment and 5 treatments (including one positive and one negative control) per block with 6 blocks.  The treatments are applied two days before, the same day and two days after the pathogen.  The plants are inoculated in two locations on the stem and each site is rated for lesion length in millimeters at 7, 14 and 21 days after inoculation with the pathogen.   I also collected stem diameter on day 7 only and recorded stem softness for each lesion on each rating day.   The blocks and treatments were randomized.  The data doesn't pass the test for normality.  I have done a double multivariate repeated measure analysis by using PROC GLM and I have also averaged the lesion lengths by day and ran the data through a PROC GLM with repeated measures.  I think I should use a nonparametric analysis because my data isn't normally distributed and have wondered if the lesion lengths for each site on an EU are covariates.  I would really appreciate some help.      

SAS codes for the multivariate repeat measures analysis;

Proc glm data = SIB;

class block trt;

model t7m1 t14m1 t21m1

           t7m2 t14m2 t21m2=trt / nouni;

repeated response 2 identity, time 3;

run;

[gcjfernandez · Accepted solution]

Please refer the following document from SAS documentation https://go.documentation.sas.com/doc/en/statug/15.2/statug_glm_examples09.htm

on Analyzing a Doubly Multivariate Repeated Measures Design. Also, you may also want to include the Block factor in the model statement.

[JA57]

Thank you. Would you recommend including block as a random factor?

[gcjfernandez]

If you use GLM then treat Block as Fixed effects. If Fixed effects for BLK is not appropriate then you need to switch to PROC MIXED.