4
Table 1. Comparison of Collaborative and PT Studies
Property
Collaborative
PT
Purpose
Measure method
variance components
and recovery bias, and to
show equivalency to a
reference method or
meet performance
requirements
Measure collaborator
result compared to
others
Method procedure
Controlled
Variants possible
Test portions
Randomized
Randomize
Levels of concentration of analyte
Full range of interest
Single level, nominal
Matrices
Multiple
Single
Disclosure
Full
Simple result
Collaborator reporting
Controlled
Ad hoc
Experimental design
Controlled
Ad hoc
Reproducibility conditions
Controlled
May vary
Repeatability conditions
Controlled
May vary
Time element
Cross-sectional
Learning curve
Cost
High
Low to moderate
Suspicious data
Infrequent
Common
Interpretability
Usually clear
Quizzical
Here we propose that the optimal solution to this issue is to divide a traditional collaborative into
separate incremental experiments (‘modules’) that preserve the randomization and control of the
planned collaborative study, but reduce the involvement and deployment load to that of a PT
study. Such an incremental collaborative study (as opposed to a cross-sectional collaborative
study) would have all of the advantages of the traditional collaborative study and of the PT study,
with none of the disadvantages of either.
Recommended to OMB by Committee on Statistics: 07-17-2013
Reviewed and approved by OMB: 07-18-2013
13
44
