PHARM/PCS V5 · Procedure 34 ANOVA and Comparisons
Analysis of Variance II: Two-Way, Single Observation
Test a treatment effect and a blocking effect together, with one observation for each combination, so the blocking absorbs nuisance variation.
What it computes
Two-factor ANOVA with one observation per cell. Each "file" is a treatment; rows within each file are blocks. Tests for significant differences between treatments and between blocks.
Inputs and outputs
You provide
- One block of values per treatment, one value per block (row); every treatment needs the same number of rows
PHARM/PCS reports
- Sums of squares for treatments, blocks, and error
- F ratios for treatments and for blocks, each judged at the 95% and 99% levels
- A profile plot of the values by treatment
When to use it
When two factors act at once (typically a treatment factor plus a blocking factor such as litters, days, or subjects) and you have exactly one observation per combination. Blocking absorbs nuisance variation that would otherwise inflate the error term in a one-way analysis (Procedure 33).
Reading the results
With a single observation per cell you cannot test interaction; the design assumes treatment and block effects simply add. If the two factors genuinely interact, this model misleads and you need replication (Procedure 35). Otherwise follow a significant treatment F with the usual post-hoc tests (Procedures 36-38, 44).
Faithful to the original
Results match the 1986 program to every printed digit. Both F ratios are compared against critical values computed directly rather than read from a table.
This page describes the procedure. Calculations run inside the licensed PHARM/PCS V5 application. PHARM/PCS is a trademark of MCS Scientific LLC.