From a SAS example with variable dose at 4 levels :

proc glm;

class dose;

model y=dose;

contrast 'linear' dose -3 -1 1 3;

run;

I don't know how the contrast -3 -1 1 3 has been derived and so I can't derive my own contrast statement for my data with 11 dose levels. I am used to determining contrast statements by writing as below. In the example below I will assume I want to test :

Ho : mu11 - mu12 = 0

mu11 - mu12 = (mu+a1+b1+ab11) - (mu+a1+b2+ab12) = b1 - b2 + ab11 + ab12

I would then write out the contrast as contrast 'ab11 - ab12' b 1 -1;

Here for the linear trend I'm not sure how to express the contrast in terms of model parameters. Can anyone help. I guess if I could understand the example given above this would help me. Thanks. ]]>

I'm doing proteomics and I work with drought stress. I wanted to study two different fractions: the soluble (sol) proteins and the microsomal (mic) proteins and three different drought levels: control (C), moderate drought (MD) and severe drought(SD), so I got 6 different conditions in total. Also, for 4 conditions I got 5 replicates but for C-mic I got 10 and for MD-mic I got 8. I got around 400 proteins in total, so my excel looks like the one I attached (but with only 11 proteins)

I want to know if each protein changes from control conditions to moderate drought and/or to severe drought, and also know which of those that change are the most responsible for the differences between treatments.

*My data values are Spectral Counts, which I divided by the protein molecular weight and x100. Then I performed a log2 to make the sampels more normal.

To test for normality of each condition I performed a Saphiro-wilk test, and to test for difference sin variances a fisher test, both using excel (saphiro wilk with RealStats pack for excel).

then I performed a type2 t student test using excel again, one test per protein comparing Cmic with MDmic, Csol with MDsol, Cmic with SDmic and Csol with SDsol. For the cases in which the variances were different I performed a type 3 t student test.

I dont know if this is the most suitable way to do this, I was checking the ONE WAY and TWO WAY anova tests, but it would just tell me that the menas of CMm-MDm-SDm are different, right?and to know where that difference is I should perform a t test...am I right?

I didnt only do statistics but calculated the fold change, so combining LFC with t student I selected the proteins that significantly changed.

Ok, assuming I am right about what I said above now that I know which proteins are statistically different I want to know which of them are responsible for the differences between the Cmic samples-MDmic samples and SDmic samples, and also the same for the sol samples. I thought of doing a PCA analysis. I guess I cant do that in excel and should use SPSS, but my license is not working anymore (sigh), maybe with sigmaplot? Anyways, I guess that to make it more accurate I first pperform a PCA entering all the protein data for Cmic and for MDmic. That should give me the loading, which are the proteins that affect the separation of Cmic and MDmic the most? What about an ICA?

Also...many sampels had different variances, can I then perform a PCA/ICA?

Also, as you see, I treated the different protein fractions (mic and sol) separately but maybe it would also be interesting to do statistics using that variable as well (not drought level only, but fraction too). Well there I am even more lost!

Can anybody help me please? Thank you so so much really

The experiment is investigating whether IQ changes based on certain training (there are 2 courses and a control - nothing). I have the IQ of individuals before and after the courses and I want to know if the IQ will increase as a result of the courses, but not as a result of receiving no training.

The paired t test would require separating the data into groups based on the course, this would show if the IQs changed significantly within the groups but wouldn't show if the groups were significant from each other... Therefore I was thinking about the two sample anova. Is the two sample ANOVA the way to go?

I am also going to illustrate any changes in the IQs for each group, by calculating the differences from before and after. Would this calculation be something which I could statistically test? carrying out an ANOVA between the difference and the courses?

Thanks for taking the time to look at this for me!!

Hopefully its a lot simpler than I am making it!! ]]>