ANOVA Help please !!!

[Gamewizard]

Hi, I need help with my ANOVA question. There are some things which I am having trouble understanding. Sorry this is going to be long but i need to include details so that it is understandable, i'll try to keep it short though.

 

Below is the data I am working on.

 

Aspirin

5.1

3.8

4.9

2.7

4.7

5.3

3.8

4.0

 

Paracetamol

4.3

3.8

3.7

4.8

3.5

4.1

3.4

4.0

 

Ibuprophen

4.2

4.1

4.9

2.9

3.2

5.2

3.0

2.7

 

Placebo

5.6

5.2

5.3

4.2

5.5

6.0

6.3

5.6

 

So far, I have calculated the treatment totals (tj) calculated crude sums & sums of squares:

SUMx 34.3+31.6+30.2+43.7= 139.8

SUMx^2 5.1^2+3.8^2 etc = 640.52

SUMTj^2 34.3^2+31.6^2+30.2^2+43.7^2= 4996.78

 

Then sums of squares are calculated:

 

SStot = sumx^2-(sumx^2)/rc = 640.52-(139.8^2)/32 = 29.76875

 

SStr= sumtj^2/r-(sumx)^2/rc = 4996.78/8-610.75125 = 13.84625

 

SSres= sstot-sstr = 29.7685-13.84625 = 15.92250

 

Then next step is to construct ANOVA table:

 

Source of variation SS Df MS MSR

Tablets residual 13.84625 3 4.61542 8.12

15.92250 28 0.56866

Total 29.76875 31

 

 

 

This is all correct and I am happy with it but however, i dont know what to do after this and the final answer i got from the teacher was F28^3(0.001)= 7.05 < 8.12 I dont understand how we get to this answer

[alpha2cen]

Simple example. There are three groups we'd like to compare.

A(1,000,000member) B(1,000,000member) C(1,000,000member)

Because of many reason (economical, time, ...), we do sampling some of them. Your example is 4x8.

When we compare A, B, C difference, the difference A B C effect should be bigger than A B C inside error, at this time we can say there is a difference between A, B and C .

The comparing scale is F factor.

If F factor is big, ABC main effect is bigger than ABC inside error.

If F factor is small, ABC main effect is smaller than ABC inside error.

Your computed result 8.12 is bigger than inside error.

So the difference between good drug and poor drug exists within 0.00l critical value.

Edited December 31, 2010 by alpha2cen