I’m not an I/O psychologist (and I don’t even play one on TV), but I work with some of the most respected, Ph.D.-trained I/Os in the country. I focus most of my attention on how our tools help our healthcare clients meet their goals, but occasionally I get to sit in on assessment development meetings with our consultants and researchers and get into the nuts and bolts of assessment. It’s fascinating (if you are into psychoanalytic research and statistics, I suppose) to see them at work, creating or tweaking assessments based on thousands of data points to ensure that we maximize their predictive nature.
This made me think it might be useful to give you a peak behind Oz’s curtain, so to speak, to better understand how a test predicts performance. Just like with the wonderful Oz, it’s not really magic but neither is it smoke and mirrors. As complicated as it can be, sometimes, it’s pretty simple. Our tools are predictive and they are well received by candidates for one reason: We simply ask people how they tend to behave across time, setting, and situation. By doing so, candidates divulge information that either shows them as a good fit or as someone who will be toxic in the environment and ultimately turnover.
A typical screening assessment might consist of 120 questions (“items”), but let’s consider two that are used in some of these tools and how they predict turnover:
1) “I never show up late for work.”
Employees who leave their jobs at our hospital-clients disagree with this item significantly more often than people who stay at their jobs.
2) “How many unscheduled or unexcused absences per year do you think is acceptable?"
People who stay at their jobs our hospital-clients indicate a significantly lower number of unscheduled/unexcused absences are acceptable relative to those who leave their jobs. Further, those who end up being terminated due to job abandonment indicate a much higher number of acceptable unexcused absences.
These items really are simple and a healthcare hiring assessment can predict future behavior because candidates tell us how they have behaved in similar environments in the past. Now expand this concept to hundreds of items that our research shows are predictive of everything from a person’s level of compassion, degree of adaptability, ability to innovate, or likelihood to be collaborative and patient focused and you have some powerful data.
This sort of data, though, also points out the need to understand what you want/need to measure. It’s why an off-the-shelf test may not work and, in fact, make it less likely that you’ll select for the attributes you seek.
You need some expert help to define what you are looking for and to make sure you have the right tool/assessment, properly configured to meet your needs. To learn more, see our whitepaper below: