Legal Defensibility
As an assessment provider, one of our greatest priorities is to ensure the legal defensibility of our assessments. Legal defensibility is understandably a concern for organizations looking to incorporate assessments into their people management processes. However, when an organization uses assessments that are professionally developed and well-validated, assessments actually enhance the objectivity, equitability, and legal defensibility of your decisions.
Through our test development efforts, we ensure the legal defensibility of our assessments in several ways:
Validating that the assessments are predictive of job performance
Our goal is to help organizations make better talent decisions, and this is only possible if the assessments are validated to predict job performance. We specifically develop our assessments to optimize prediction of job performance through rigorous validation studies conducted by our team of I/O psychologists. Even after a test is launched, we continue to perform hundreds of additional local validation studies with our client organizations to further optimize their predictiveness across an extensive range of positions, industries, and job performance measures.
Adhering to governmental standards for accuracy and defensibility
Criteria’s assessments exceed the highest government thresholds for accuracy and defensibility. (Learn more about current standards in Testing & Assessment: An Employer’s Guide to Good Practices published by the United States Department of Labor in 2000). Our test development, validation, and implementations adhere to all relevant government and professional guidelines (e.g., The Uniform Guidelines on Employee Selection Procedures). The Uniform Guidelines (UGESP), for example, explicitly recognizes the right of employers to use pre-employment tests to make hiring decisions as long as those tests are job-related. We follow these guidelines closely, both when developing assessments and when recommending them to our clients.
Independent Bias Audit of Criteria's Assessment Platform
An independent audit of Criteria’s assessments undertaken in June 2024 found no practically significant disparate impact1 against any gender, any US racial category, or against any combination of gender and US racial category. The 2024 bias audit used the data and output standards required by NYC Local Law 144, and examined the results obtained by job candidates who completed Criteria assessments over the previous five years (encompassing over 551,022 candidates). The detailed findings of the bias audit results are used by Criteria to guide the development of our assessments, recommended test batteries and suggested score ranges.
Targeting our development efforts to minimize subgroup differences
All of Criteria’s test development efforts are designed to ensure that our tests are non-discriminatory and that subgroup differences are minimized/eliminated, while maintaining the accuracy and utility of the assessment. We do this by reviewing all test content for cultural sensitivity and for differential item functioning to ensure the test is free of content bias across cultural and gender groups.
Taking a thoughtful, transparent, and ethical approach to AI
Criteria’s approach to AI is backed by the same rigorous scientific standards that we uphold for our traditional products. These high standards enable us to deliver solutions that safeguard the privacy and security of our candidate and client data while remaining fully compliant with the latest AI legislation and best-practice guidelines. Learn more about Criteria's approach to AI.
Working side-by-side with our clients to ensure defensible implementations
We partner with our customers to help them design an assessment strategy that is job-related and predictive of performance. Our priority is to help organizations make better people decisions, and this goes hand-in-hand with implementing a predictive and legally defensible process that is specific to your organization’s needs.
___________
1 Practical significance refers to the US EEOC 4/5ths rule, which requires that the selection rate for any one group must be within 80% (or 4/5ths) of the selection rate for the most selected group.