Statistical text

[Olanike]

Hi All,

I will be glad is someone can kindly help me out with this.

I have a split plot experiment based on randomized complete block design (RCBD) with two factors (crop rotations and tillage practices) and four replications. I analyzed my results using PROC MIXED (but with no interactions, that was the way I wanted it).

My code for multi year analysis

proc mixed data= Data;

class Year Rep T_System;

model &var = T_System;

random Rep(year)year year*T_System/solution;

repeated year/subject=rep*T_System type=ar(1) RCORR;

Lsmeans T_System/diff=all;

I wrote my statistical text as:

"The results from each year were analyzed separately with treatments (tillage and rotations) and replication treated as fixed effect and random effect, respectively. Combined analysis was then carried out using repeated measures and year was assumed as a random effect. Differences among treatments were tested using the Tukey HSD test".

But someone commented that

My text appeared as if tillage and rotation were considered as independent factors, and if that had been the case I should have analyzed them as such, with interactions, etc.

Please how better do I construct my statistical text

Thanks

-Ola

[sld]

I presume that T_System has four levels: the 2x2 factorial comprised of crop rotation and tillage practice. I think it is more straightforward and intuitive to analyze the 4 treatment combinations as a 2x2 factorial, allowing for a default assessment of the main effect of crop rotation, the main effect of tillage practice, and their interaction. Choosing to ignore the interaction could be hard to justify; and if you are doing all pairwise comparisons among the 4 combinations, then you are implicitly allowing the possibility of interaction.

[Olanike]

Hi Sld,

Thanks for the response. I did not do any interaction but did the contrasts of all the crops as below:

estimate 'High_LowSG_M' T_System 0 0 0 0 0 1 0 0 0 0 0 -1;

estimate 'High_LowDG_M' T_System 0 1 0 0 0 0 0 -1 0 0 0 0;

estimate 'High_LowSC_W' T_System 0 0 0 0 1 0 0 0 0 0 -1 0;

estimate 'High_LowDC_W' T_System 1 0 0 0 0 0 -1 0 0 0 0 0;

estimate 'High_LowDlen' T_System 0 0 1 0 0 0 0 0 -1 0 0 0;

estimate 'High_LowDmus' T_System 0 0 0 1 0 0 0 0 0 -1 0 0;

Thanks.

[Olanike]

Thanks for the response Sld. I did not do any interaction but did the contrasts of all the crops as below:

estimate 'High_LowSG_M' T_System 0 0 0 0 0 1 0 0 0 0 0 -1;

estimate 'High_LowDG_M' T_System 0 1 0 0 0 0 0 -1 0 0 0 0;

estimate 'High_LowSC_W' T_System 0 0 0 0 1 0 0 0 0 0 -1 0;

estimate 'High_LowDC_W' T_System 1 0 0 0 0 0 -1 0 0 0 0 0;

estimate 'High_LowDlen' T_System 0 0 1 0 0 0 0 0 -1 0 0 0;

estimate 'High_LowDmus' T_System 0 0 0 1 0 0 0 0 0 -1 0 0;

Thanks.

[sld]

The ESTIMATE statements imply that T_System has 12 levels. (I wrongly assumed 4, but that wasn't based on anything that you wrote.) I don't know what these 12 levels are, so I can't provide any feedback.