For hourly workers (Unicru case), perhaps there’s not enough info out there (or the info might be inaccurate) so tests would help. However, for traditional employers, I don’t think these tests are as useful.

When employing salaried employees, I’ve seen companies give simple coding tasks for programmers and college level math problems for insurance companies. One difficulty with an over-reliance of this data is that you’ll bias your hiring on those who are good at these tests (much the same way that traditionally people over-hire people who are good at interviewing). The other issue is that there’s a ton of information already that people generally put little weight on (GPA, test scores — for the insurance company which gave the math test, looking at their success in math classes could be a reasonable proxy for that).

I agree that 3 hours of interviews + a resume isn’t sufficient towards making a fully-qualified decision, so any additional information is helpful. Instead of additional tests/questionnaires, though, I think HR is better served seeking information themselves, which some have done in the form of web searches for individuals and other information-gathering.

Basically, the hiring process can be optimized in a manner similar to what professional sports teams do (gather as much information as possible and then try to extrapolate one’s career to past athletes who followed the same trajectory). The aspect which I think best compares to these questionnaires is the combines, where how fast someone runs a 40 meter dash on one day, a day where either the person could be suffering from jet lag, a cold, unfamiliar surroundings, sore muscles, or anything else, could affect whether he’s a first rounder or a late second round draft choice…an entire 4 year body of work is basically given the same weight as what happens in <5 seconds.

More information is always good – but I wonder if the key is actually better analysis of the existing information, extracting the fact that success as an actuary has a high correlation to performance in probability courses, which makes sense, but maybe the models may also determine that successful consultants also tend to have done well in literature classes (something less obvious).

Not so sure about using questionaire data to model and predict hiring decisions, but I do think that data from past performance and work history will be used to better match people and jobs.

Another interesting view into the way data will affect our future is by looking at our kids. I teach part time at a local Middle School. these kids have multiple information sources they interact with concurrently, through multiple devices. An impact of this behavior on modern application design could be a move away applications delivering a single simple function, say a spreadsheet. Every function needs to be available everywhere and at any time.