Hi all,
Please consider the following problem and advise me.
A potency assay.
There are four antigens (drugs) each at two levels (0.02mg and 0.1 mg)
We, therefore, have eight treatments (T1…T8).
The purpose of the assay is to assess the potency of the antigens.
One is a control antigen (whose potency is known) against which the three
others will be compared.
The eight treatments are assigned to one of eight injection positions on an
animal.
These eight injection positions consist of four sites on left hand side of
animal and four sites on right hand side of animal. Each site is opposite
each other on the animal as shown
Animal 1
Site
Side 1 2 3 4
LHS T1 T4 T6 T8
RHS T2 T7 T3 T5
There are eight injection positions LHS1,LHS2, LHS3,LHS4, RHS1,RHS2,
RHS3,RHS4. Each of the eight treatments was applied to each possible
position.
i.e. there are eight animals, each of the eight treatments appears at each
of the injection positions.
For example, say for T1
Animal1 T1 LHS1
Animal2 T1 LHS2
Animal3 T1 LHS3
Animal4 T1 LHS4
Animal5 T1 RHS1
Animal6 T1 RHS2
Animal7 T1 RHS3
Animal1 T1 RHS1
There are eight animals in a group and three groups on which the experiment
is replicated.
The response variable a measure of the reaction of each treatment, at each
side x site position on each animal.
The primary purpose of the experiment is to estimate the potency of the
antigens.
Now, the expt is laid out as a Latin Square, eight treatments (antigen x
level structure), eight animals (assumed to be blocks) and eight positions
(side x site structure...also a block). There are three Squares.
Is it possible to use SPSS to carry out the analysis and if so how best to
do it.
Presumably its some extension of Latin Square treating the factors as fixed
effects.
Alternatively I can see how it could be modelled by a random effects model
treating animal as a random effect, how would this be best done in SPSS.
My problem is not really the parametrisation of the model, though I would
appreciate any input on this, to confirm my own interpretation. The problem
is more how to get SPSS to analyse the assay for me in such a way that I
get estimates for the four antigens.
|