# Internal Revenue Service **Entity class:** Federal tax agency ## Definition and history The Internal Revenue Service introduced the Discriminant Function system in 1969 to rank returns for possible examination. [[articles/Intelligence Inside|Intelligence Inside]] reads this as an early example of supervised scoring directing national institutional attention. The agency also operated research examination programs that supplied statistically designed data outside the ordinary scored-selection stream. Human examiners and classifiers remained part of the resulting decision process. ## Mechanism and limits A return-scoring model allocates scarce examination capacity according to estimated yield. Its own choices affect which outcomes become visible, so research samples from TCMP and NRP have a separate calibration role. The distinction among model score, human selection, examination, and later research sample keeps the administrative process legible. ## Relationships **Operates:** [[wiki/IRS Discriminant Function|IRS Discriminant Function]]. **Research samples:** [[wiki/Taxpayer Compliance Measurement Program|Taxpayer Compliance Measurement Program]] and [[wiki/National Research Program|National Research Program]]. **Modeling issue:** [[wiki/Selective Labels|Selective Labels]]. **Article route:** [[articles/Intelligence Inside|Intelligence Inside]]. ## Sources [IRS Research Bulletin, DIF model history](https://www.stayexempt.irs.gov/pub/irs-prior/p1500--2022.pdf); [[articles/Intelligence Inside|Intelligence Inside]].