Skip to content
Universal Model ProtocolDocs

References

§26NormativeDraft 2.3Latest draft

26.1 Normative references

Section titled “26.1 Normative references”

An implementation needs these documents to conform. Where a version or date is given, that version applies.

26.2 Informative references

Section titled “26.2 Informative references”

These references explain or motivate the rules; an implementation does not need them to conform. Most are carried from the research review. The Checked column marks which entries were re-verified for this draft.

Standards and guidance

ReferenceChecked
Open Inference Protocol (KServe V2)no
Open Policy Agent and Regono
SR 26-2 (April 2026), superseding SR 11-7no

Methods

ReferenceChecked
Talts et al. (2018), simulation-based calibration, arXiv:1804.06788no
Vehtari, Simpson, Gelman, Yao and Gabry (2024), Pareto smoothed importance sampling, JMLR 25(72)no
Vehtari, Gelman, Simpson, Carpenter and Bürkner (2021), rank-normalisation and improved R-hat and MCSE, Bayesian Analysis 16(2)no
Jin et al. (2017), Bayesian MMM with carryover and shape effects; Chan and Perry (2017), MMM challengesno
Runge et al. (2024), Robyn calibration objectives, arXiv:2403.14674no
Endogenous-spend MMM benchmark (2026), arXiv:2608.21130no
Nemhauser, Wolsey and Fisher (1978), greedy submodular maximisationno
Khuller, Moss and Naor (1999), the budgeted maximum coverage problemno
Leskovec et al. (2007), cost-effective outbreak detection (cost-scaled greedy with a best-single fallback)no
Bertrand, Duflo and Mullainathan (2004), serial correlation in DiDno
Chen and Au (2022), Trimmed Match, Annals of Applied Statistics 16(1)no
Braun and Schwartz (2025), divergent delivery, Journal of Marketing 89(2)no
Burtch et al. (2025), Lift versus A/B tests, arXiv:2508.21251no
Smith and Winkler (2006), the optimiser’s curseno
Box (1980), prior predictive checking; Neuenschwander et al. (2010), MAP priors; Schmidli et al. (2014), robust mixture priorsno
Benjamini and Hochberg (1995), controlling the false discovery rateno
Benjamini and Yekutieli (2001), false discovery rate under dependencyno
Holm (1979), a simple sequentially rejective multiple test procedureno
DerSimonian and Laird (1986), random-effects meta-analysisno
IntHout et al. (2014), the Hartung-Knapp-Sidik-Jonkman method for random-effects meta-analysisno
Egger et al. (1997), bias in meta-analysis detected by a graphical testno
Dwork et al. (2006), calibrating noise to sensitivity in private data analysisno
Fieller (1954), interval estimation of a rationo
Hyndman and Fan (1996), sample quantiles in statistical packagesno
Morris, White and Crowther (2019), using simulation studies to evaluate statistical methodsno
Raiffa and Schlaifer (1961), applied statistical decision theoryno
O’Hagan et al. (2006), Uncertain Judgements: Eliciting Experts’ Probabilities, Wileyno
Lakens et al. (2018), equivalence testingno
Embrechts et al. (2013), model uncertainty and VaR aggregation (dependence bounds)no
Pimentel, Murta, Braganholo and Freire (2019), notebook reproducibility, MSRno
Lei, Yu et al. (2024), Spider 2.0, arXiv:2411.07763; Deng et al., ReFoRCE, arXiv:2502.00675no
Greshake et al. (2023), indirect prompt injection, AISecno
Debenedetti et al. (2025), defeating prompt injections by designno
Stinnett and Mullahy (1998), net health benefitno

Software. PyMC-Marketing (add_lift_test_measurements), Google Meridian (CalibrationBuilder, ROI priors), Meta Robyn (calibration_scope), ArviZ InferenceData, Trillian, Sigstore Rekor.

26.3 Provenance

Section titled “26.3 Provenance”

This specification grew out of these project documents. They are informative background only (section 0): UMP Standard Document Plan; Model Governance Framework for Internal Models; MMM to EMA Technical Guide (v2, with non-geo experimentation); modeling-api Reference Notes (v0.15); UMP Decision Stack site; UMP Decision Stack Research Review.