PAC's assessment was built from the ground up and utilizes the most recent advances in IO Psychology Research to ensure the highest reliability and predictive ability.
Our assessment is based on the US Department of Labor’s O*Net Content Model. The model comprehensively covers key dimensions related to job performance, reflecting the state of scientific knowledge in multiple content areas (e.g., personality, needs, skills, and context). The model was recently reviewed/supported by the National Academy of Sciences (2010).
We developed scales to measure each facet in the O*Net content model. Our assessment was developed by Rhys Lewis, Ph.D. – an Industrial/Organizational Psychologists specializing in employee assessment design and validation
Reliability is the overall consistency of a measure. Reliability is usually evaluated using Cronbach's α (alpha), a metric that reflects the consistency of results across items within a scale. In published assessments, a Cronbach's α of .7 is considered minimally acceptable, .8 is considered good, and .9 is considered excellent. Ideally, all scales on an assessment should have a Cronbach's α of .8 or higher. In real life, few assessments meet that goal.
All of PAC's scales have Cronbach's α above .8. The average reliability is .88, and the minimum is .82. This level of reliability suggests that PAC's assessment is suitable for use in both personal and high-stakes testing.
Sometimes a lack of difference is desirable. To promote equal employment goals, it is helpful when protected groups score similarly to other groups. PAC has explored the differences between groups on three grounds: ethnicity, gender, and age.
Group differences are measured using Cohen’s d (Cohen, 1988), which takes into account both the size and consistency of the difference between groups. When interpreting Cohen’s d, .2 is considered a small effect, .5 is a medium effect, and .8 is a large effect.
Taken together, ethnic, age, and gender differences are generally small or expected, suggesting that PAC's assessment is appropriate for use across protected groups.
PAC predicts fit with careers based on a person's interests, personality, needs, and preferences. Predictive validation asks the question "How well does it work? In other words, to what degree can PAC predict a person's reactions to careers?