Researchers at the Oregon Health Sciences University conducted an experiment
to study how delivery of brain cancer antibody is influenced by:

 -- tumor size (3 levels)
 -- antibody molecular weight (3 levels)
 -- blood/brain barrier disruption (2 levels)
 -- delivery route (2 levels)

72 female rats of approximately the same age and size were randomly assigned
to the 36 treatment combinations (2 rats per treatment). Tumor size was varied
by starting treatments either 8, 12, or 16 days after innoculation with tumor cells.
Three antibodies (AIB, MTX, DEX7) were used which have very different molecular
weights. Disruption (BD) occurred by using an intra-arterial injection of 
mannitol to one side of the brain and was compared to a similar injection of saline
(NS). The antibody for each rat was delivered by either intra-arterial (IA) or 
intra-venous (IV) injection.

The outcomes measured were the concentrations of the antibody in the brain 
(counts of radio-tagged molecules) around 
the tumor (BAT) and in the other, lateral half of the brain (LH). The ratio of
these two quantities is considered a key measure of antibody delivery.

Data are coded as:

   weight:  1=AIB, 2=MTX, 3=DEX7
   route:   1=IA, 2=IV
   disrupt: 1=BD, 2=NS

Data are from:

Barnett PA et al. (1995). "Differential permeability and quantitative MR imaging
of a human lung carcinoma brain xenograft in the nude rat," Journal of Pathology,
146: 436-449.

These data were reproduced in the book:

Ramsey and Schafer (1997). The Statistical Sleuth: A course in methods 
of data analysis, Duxbury Press.