SELECTsd.ID,

sd.string,

sd.category,

ce.experiment

FROM @SampleData sd

INNER JOIN (

SELECT category,

CASEWHEN category % 7 = 0 THEN 'A'

WHEN category % 3 = 0 THEN 'B'

ELSE 'C'

END AS experiment

FROM @SampleData

) ce

ON sd.category = ce.category

ORDER BY ce.experiment, sd.string

Following on from Dwain's comments, another issue is non-atomic data, category is not just a primary key, it also denotes the experiment. If you have a CategoryExperiment relation, then it's easy street.