Testing Validation

We extensively validate our assessments, both in the process of developing the assessments as well as on an ongoing basis with client implementations. We adhere to relevant government and professional guidelines (e.g., UGESP), and take a multilayered approach that optimizes prediction of job performance, while mitigating any subgroup differences.

As part of this process, we look to enhance the reliability of our assessments (e.g., Cronbach’s Alpha), as well as construct validity, and criterion-related validity. On skill-based assessments and those where the assessment itself is considered part of the job domain, we also engage in content validation. The following describes each of these at a high level:

Reliability

Reliability refers to the consistency of an assessment’s scores. We evaluate it in several ways, including internal consistency (such as Cronbach’s alpha), test-retest reliability (are scores stable over time?), and, where applicable, alternate-forms and inter-rater reliability. 

Construct Validity

This determines if a test is measuring what it is supposed to measure. Depending on the test in question, the process for determining construct validity typically involves leveraging other scientifically validated tests to find consistent correlations (e.g., convergent, as well as discriminant validity).

Criterion Validity

Also known as predictive validity, this refers to the ability of the test to correlate with a business outcome. For example, is the test correlated with lower turnover, stronger performance metrics, or higher sales? This validation step involves gathering data from organizations to demonstrate that the assessments can predict these outcomes. We also continue to validate our assessments with local validation studies and case studies performed with our customers.