Statistical analysis may be used at the class-certification stage to evaluate factual questions concerning a proposed class. Depending on the matter, an analysis may help identify what can be measured from available records, describe patterns or variation in the data, evaluate information obtained from a sample, or examine statistical relationships among relevant variables.
Statistical analysis alone does not determine whether a class should be certified. That determination is for the court under the applicable legal standards. The appropriate analytical methodology, if any, depends on the particular questions presented, the available evidence, the quality and completeness of the underlying information, and the assumptions required for the analysis.
Below are four analytical approaches that may be considered at the class-certification stage, depending on the circumstances.
Analyze Historical Records and Data
Where relevant historical data are available, these materials may provide a starting point for analysis.
In employment cases, for example, potentially relevant records may include timekeeping, payroll, scheduling, point-of-sale, electronic access, or other operational data. In other matters, relevant records might include transactions, customer accounts, insurance claims, pricing information, or other business records.
Depending on the circumstances, an analysis may encompass an entire available population or a subset of records selected using an appropriate sampling methodology.
Historical records can be analyzed to determine what information they contain and what measures can be constructed from them. An analysis might examine the frequency of a recorded event or whether observable patterns vary across employees, locations, time periods, job positions, transaction types, or other identifiable groups.
Such an analysis may also identify limitations in the available records, including information that is missing, incomplete, or not captured in the data.
Propose a Data and Document Roadmap
In some matters, potentially relevant records may be identified before they have been produced or analyzed. An analytical roadmap can identify potential data sources, describe how records might be linked, and identify measurements that could be considered.
In those circumstances, an expert may describe the types of records that could be relevant and the analyses that potentially could be performed if appropriate data become available.
For example, timekeeping records may identify when an employee clocked in but may not contain information about events occurring before that time. Depending on the allegations, other records — such as scheduling, electronic access, computer login, dispatch, or security data — may contain additional relevant information.
Analyze Sampled Class Member Evidence
Some factual questions may involve information that is not contained in historical business records. In appropriate circumstances, information may instead be collected from a sample of class members through surveys, testimony, interviews, declarations, or other methods.
Whether results obtained from sampled individuals can appropriately be generalized to a broader population depends on the methodology used to select the sample and collect the information. Relevant considerations may include the sampling design, sample size, response patterns and potential non-response bias, question design, and consistency of the data-collection process.
Sampled evidence also may be considered together with historical records when the two sources provide different types of information. Whether and how those sources can appropriately be combined depends on the particular methodology and circumstances.
Regression Analysis and Statistical Modeling
Regression analysis and other statistical models may be used to examine relationships among measurable characteristics in a dataset.
For example, regression analysis may evaluate whether an observed outcome is statistically associated with location, job position, supervisor, shift characteristics, transaction type, time period, or other available variables.
The interpretation of a regression analysis depends on the question being studied, the available variables, the quality of the underlying data, the model specification, and the assumptions employed. Depending on its design and the assumptions employed, regression analysis may be used to evaluate statistical associations or, in appropriate circumstances, questions involving causal relationships.
Choosing an Analytical Approach
The four approaches described above are not mutually exclusive, and not every approach will be appropriate in every matter. Historical records, additional data sources, sampled evidence, and statistical models may be used separately or in combination depending on the questions presented and the available evidence.
Ultimately, the methodology should be tailored to the particular analytical question. A statistical analysis should identify what was measured, the information on which the analysis relies, the assumptions employed, relevant limitations, and the scope of the conclusions that can reasonably be drawn.
Get Related Sources
FAQs
What types of data may be analyzed at the class-certification stage?
The relevant data depend on the allegations and the questions being evaluated. In employment matters, potentially relevant records may include timekeeping, payroll, scheduling, point-of-sale, electronic access, and other operational data. In other matters, relevant information might include transaction, customer account, insurance claim, pricing, or other business records.
Can an analysis be proposed before all potentially relevant data have been produced?
Potentially. An analytical roadmap may identify data sources that could be relevant, describe how records might be linked, and identify measurements that could be considered.
When can sampled class member evidence be generalized to a broader population?
Whether results obtained from sampled individuals can appropriately be generalized depends on the methodology used to select the sample and collect the information. Relevant considerations may include the sampling design, sample size, response rate, non-response, question design, and consistency of the data-collection process.
What can regression analysis show in a class-certification analysis?
Regression analysis can be used to examine statistical relationships among measurable characteristics in a dataset. Its interpretation depends on the question being studied, the available variables, the quality of the underlying data, the model specification, and the assumptions employed. Depending on its design and the assumptions employed, regression analysis may also be used, in appropriate circumstances, to address questions involving causal relationships.
This article is provided for informational purposes only and reflects general observations regarding wage-and-hour damages analyses. It is not intended to provide legal advice or to express an opinion regarding the merits of any particular claim or defense. Because wage-and-hour laws vary by jurisdiction and continue to evolve, readers should consult qualified legal counsel regarding the application of these laws to any specific facts or circumstances.