Statistics, Sampling & Econometrics

Statistical sampling withstands Daubert challenges when it satisfies the same criteria applied to any expert methodology. The approach must be testable, peer-reviewed, or otherwise accepted within the respective scientific community, and it must be applied with an acceptable error rate. Courts closely scrutinize both sample selection and methodology. Sampling analyses that exhibit selection bias, inadequate stratification, or a failure to account for variability within subgroups may be excluded.

IMS designs and documents sampling methodologies to withstand Daubert scrutiny, ensuring that expert opinions remain defensible under cross-examination.

Quantitative analysis is most consequential in Predominance, or the requirement that common questions outweigh individual ones. Plaintiff's representatives may retain statistical experts who use regression analysis and sampling techniques to demonstrate that a challenged policy or practice had a consistent, measurable effect across the class.

Defense experts, in turn, challenge models by introducing evidence of individualized variation, showing that putative class members experienced different circumstances or outcomes, or establishing that any apparent pattern dissolves once proper controls are applied. During class certification, disputes frequently focus on model specification and the inclusion or exclusion of key variables.

IMS experts engage in these disputes at the methodological level, stress-testing opposing models and building analyses that courts can rely on to resolve the predominance question cleanly.

In employment discrimination matters, regression analysis is used to determine whether a protected characteristic predicts an adverse outcome after controlling for explanatory factors and to quantify the magnitude of any disparity. Statistical significance indicates the finding is unlikely to be a product of random chance, but it does not tell you the disparity is legally meaningful or practically large. Properly constructed regression models must include all legitimate, non-discriminatory variables that the employer actually used in its decision-making process.

IMS statistical experts work with employment litigators to build models that are defensible both methodologically and evidentially.

Advanced data analysis and predictive modeling frequently appear in commercial disputes, where lost profits calculations, breach-of-contract damages, fraud detection, and insurance coverage are at issue. The risk of exclusion may arise if the model functions as a black box, or when the expert cannot explain, step by step, the assumptions built into the model, the training data used, and why the predictive outputs are reliable for the specific question at issue.

Courts are more receptive to data science testimony that is transparent, methodologically sound, and explained in plain language. IMS data analytics experts produce findings explainable at every layer, from raw data inputs through final conclusions, so that the opinion survives both Daubert scrutiny as well as effective cross-examination at trial.

Statistical sampling and aggregate proof methods allow mass tort parties and courts to estimate classwide harm with reliability, provided the methodology is sound. The framework for aggregate proof requires that statistical models precisely reflect the actual distribution of harm within the class and that the defendants have a meaningful opportunity to challenge the evidence.

Experts must be able to defend the representativeness of the sample and the reliability of the extrapolation to the full population. IMS statisticians are authorities in designing and executing sampling protocols specifically for high-volume tort and consumer protection matters.