Statistical Sampling and Analysis

In litigation, business negotiations, and valuations, clients are typically confronted with seemingly endless amounts of data. Statistical samples, if properly designed and gathered, can provide important and defensible information about large populations of people, products, buildings, or events. Statistical samples allow experts to make precise, accurate, and efficient statements about groups of things that cannot be evaluated in their entirety. Our sampling experts have helped clients extract meaningful information from large populations by selecting samples from thousands of insurance claims, millions of allegedly defective products, multiple buildings or locations, constantly varying websites, and potential class members.

Statistical sampling is a specialty within the field of statistics. Sampling expertise includes the ability to design a proper sampling procedure given the target population, the form of the data, and the precision needed, as well as knowledge of how to weight data, estimate variances, and calculate the appropriate margin of error. Lacking proper training, non-experts who select samples tend to select simple random samples when they should use more complex techniques, resulting in estimates that will be both unreliable and biased. NERA's extensive experience in sampling allows us to design and implement precise, efficient sampling plans, which in turn allow us to make precise and accurate statements about the larger population.

NERA's Statistical Sampling and Analysis group provides consulting and expert testimony on the design, implementation, and analysis of both simple and complex samples used to address issues in mass torts, product liability, insurance allocation, intellectual property, antitrust, and labor litigation. Our capabilities include:

• Selecting appropriate sample design, such as simple random sampling, dollar-based sampling, stratification, and cluster-based sampling
• Drawing samples from relevant materials (e.g., paper copies, electronic records, products in a warehouse, glassware sold in stores, etc.)
• Weighting data so that results correctly represent the target population
• Calculating precise estimates and computing appropriate confidence intervals
• Using tools such as the bootstrap method to determine the precision of complex statistics
• Sampling from dynamic populations, such as websites