Tools used to check standardized tests for bias may have flaws themselves. What’s going on? 

Test bias may still exist in standardized exams despite detection methods by the top experts in the field, and may result in inaccurate prediction of job and academic performance, says a new study in the Journal of Applied Psychology.

The study challenges the belief that there are no biases in standardize tests that could benefit certain racial or ethnic groups over others, by showing how such biases can be present but not detected.

In the study, researchers created a massive simulation, using nearly 16 million individual samples to create more than eight trillion pairs of test/outcome scores. They built bias into most samples to resemble real-world results.

Checking the tens of billions of scores with super-computing technology, they found the detection procedures used regularly – and repeatedly – missed the inserted biases.

The researchers hope these results lead to more research, a closer examination of existing tests, and possibly the creation of new ones.

What do you think? Share your thoughts in the comments section.

Tags: Indiana University Kelley School of Business, Journal of Applied Psychology, Test bias