Repeated-measures Designs

[<Jody.Biggs@csiro.au>]

G'Day S+ Users,

 

I am currently try to use a split-plot type ANOVA to analysis an unusual experimental design based on the examples given in the SPlus statistics Vol1 - Repeated Measures Design.

The experiment consists of a total of 42 subjects (ID) split between two treatments.

A maximum of 27 repeated measurement for each subject were taken.

Some subjects were removed from the experiment progressively through time. i.e. 3 per treatment per day so by the end of the experiment there remained only 3 subjects per treatment.

 

The results for the split-plot ANOVA are below.

What I would like to know is where it gets the 7 degrees of freedom for time factor (Hours)?

I realise that this term doesn't appear when there are equal number of measurements as per the repeated-measures design example in SPlus Statistics Vol1.

 

Also if anyone can suggest an alternative analysis I would be interested.

 

Thanks in advance

Jody

 

 

> summary(aov(LN.N2N ~ Treatment*Hours + Error(ID), data = "" weights = No.obs,))
Note: This analysis was done with weights
      The results below are on the weighted scale
Error: ID
                Df Sum of Sq  Mean Sq  F Value      Pr(F)
      Treatment  1  141.1602 141.1602 8.102543 0.00834138
          Hours  7  593.3629  84.7661 4.865543 0.00119483
Treatment:Hours  6  286.9992  47.8332 2.745607 0.03251294
      Residuals 27  470.3865  17.4217                   

 

Error: Within
                 Df Sum of Sq  Mean Sq  F Value    Pr(F)
          Hours  26  4438.693 170.7190 24.98999 0.000000
Treatment:Hours  26   223.098   8.5807  1.25605 0.180978
      Residuals 480  3279.117   6.8315