The great synthesis of the statistical sciences

The statistical sciences are somewhat of an archipelago with islands slowly drifting apart over time. For a long time, Psychometrics has been an island at the fringes of the archipelago until, about a decade ago, a bridge was constructed between psychometrics and statistical physics. A bridge that saw quite some traffic and led to what we now call network psychometrics. 

Over the years it became clear that bridges can be constructed between most of the islands, and that item response theory models from psychometrics are a central piece in their construction. The latest bridge is between the machine learning island and psychometrics. We hope this bridge will also see a lot of traffic, both ways, and the archipelago becomes a densely connected network of islands where insight and information move freely across the bridges from one to the other.

For the (educational) measurement community, this latest bridge provides insight into how machine learning can influence measurement in a way that does justice to the traditions and (often legal) demands put on measurement applications. If, for example, a machine can learn gender or race from responses to exam questions you at least need to know. This is what this particular bridge allows for.

For more conversation on this topic, contact us at info@metior.consulting.