When considering whether to incorporate assessments and predictive analytics into your hiring process, it is worth paying attention to the following facts

A growing body of research continues to show that valid and reliable assessment instruments, when used correctly, can improve hiring accuracy and produce a significant, measurable return on investment.

Assessment tools or approaches that compare company and position-specific success factors with important job outcomes rank as the most accurate, relevant and legally defensible systems available in the marketplace. Such approaches typically require that a template or benchmark be created for a position. Benchmarks are created by collecting competency and performance data from existing employees in the position. Complex statistical analyses are then conducted to determine which competencies best predict both successful and unsuccessful performance. Once these competencies have been identified, organizations can use their customized and relevant performance models to significantly improve their hiring accuracy.

If you currently use an assessment tool or approach that does not follow the model above, you are likely using a tool with limited predictive accuracy.

There are really several factors that should be considered. One of the most important considerations, however, is discussed below.

Ask the vendor to provide you with objective evidence detailing correlations obtained between specific instrument predictors (i.e., competencies) and important job outcomes. Ask them to provide research reports of client return on investment. If no such information is available, seriously reconsider your decision to use that tool.

Most general assessment tools available in the marketplace produce correlations ranging from 0.15 to 0.25, where 1.00 is a perfectly corresponding relationship. The closer correlations come to 1.00, the more accurate you can expect the instrument to be. Using just a resumé and interview, the chances of hiring a successful candidate is from .15 to .30. When adding industry or normed psychometric assessments, this increase to between .40 to .50.

As a point of contrast, PHD tools and our benchmarking methodology routinely achieve correlations between independent predictors and specific job outcomes of 0.70 to 0.90 or higher. Furthermore, in our published client ROI studies, the statistical differences between the hiring accuracy of a client's previous hiring procedures and the accuracy of the performance benchmarks are so large that they are often associated with probability values of p < 0.00001. This means that the results may only likely be due to chance occurence 1 time in ten thousand applications of the system. It is worth noting here that acceptibility for a significant result in science is typically any result associated with a probability value of less than p < 0.05: this means that a result may be due to chance 5 times or less in one hundred trials or applications. Our results have exceeded this acceptibility threshold by a significant margin.