AOAC Guidance on FA Immunoassay Validation (August 2023)

1444 1445 1446 1447 1448 1449 1450 1451 1452 1454 1455 1456 1457 1458 1459 1460 1453

https://cran.r-project.org/web/packages/VCA/index.html and https://cran.r-

project.org/web/packages/VCA/VCA.pdf

1.4.1.1. Model statements in R have the general form:

Response ~ terms

Where “Response” is the numeric response vector and “terms” is a series of terms indicating the predictor variables in some correct syntax dependent on the

command being used.

For VCA package in general, we will use the following two types of model

statements:

If Analyst is nested:

Result ~ Lot/Analyst/TP

If Analyst is not nested:

Result ~ (Lot+Analyst)/TP

The names used here such as “Lot,” “Analyst,” “TP” and “Result” are objects defined when the data table is read into the software, and may change depending on the

data table.

1.4.2. Example Code for 3-level ANOVA (Designs 1a and 1b):

Library(VCA)

1461

Data1<- read.csv("Test Data A1b.csv")

1462

fit1<- fitVCA(form=Result~(Lot+Analyst), Data=Data1) # Analyst not nested within Lot

1463 1464 1465 1466

fit2<- fitVCA(form=Result~Lot/Analyst, Data=Data1) # Analyst nested within Lot

1.4.3. Example Code for 4-level ANOVA (Designs 2a and 2b):

library(VCA)

1467

Data2<- read.csv("Test Data A2b.csv")

1468

fit1<- fitVCA(form=Result~(Lot+Analyst)/TP, Data=Data2) # Analyst not nested within Lot

1469 1470 1471 1472 1473 1474 1475 1476

fit2<- fitVCA(form=Result~Lot/Analyst/TP, Data=Data2) # Analyst nested within Lot

1.4.4. Note in this package in R (as with most ANOVA procedures in R) you should not include the lowest order factor in the model statement. If you do, the ANOVA table will be incorrect. It is assumed that the lowest factor will be nested. The lowest order factor will be listed in

the ANOVA Table as "error.”