I have the role in the hiring process to perform technical phone interviews with candidates. The environment requires to filter for the top 10% of the people I interview (based on experience measured in terms of technical understanding and analytic thinking), roughly 2%-5% of the applicants. I would like to reduce the number of people which I have to interview.
I think about creating a small tailored test for the positions with 5-10 questions, not to be evaluated but to give people the right impression of what we will expect, with the hint that if they don't figure out the question at all (including the use of Google), then they should consider to go for another position.
I have several problems/questions:
this could be interpreted as arrogance, how to avoid this
how to introduce something like this in the process
how strong should we try to filter before interviewing
Edit: Information which may help, and points the questions made me think about:
The hiring process has 6 stages:
eval of resume by HR
telephone interview with HR
evaluating of resume by technical person (me)
interview by manager (my boss.
telephone interview by me
My feeling is that in stage 1) we have a lot of sub-par candidates. If I would take the overall population of PHDs in a physics or engineering, i would expect that most people from the top 25% should be able to pass Step 6. I also would expect that a big majority of people does not reach step 5. At step 2, I find myself often unable to judge the quality of the applicants and i find that HR is completely in the dark for any objective criterium for what we are looking for and can not possibly interview for that.
Lets call the test "technical quiz" (as suggested) and i would like to introduce this before step 3, possibly before step 2. A typical question (equivalent problem type from my previous field of work) would be connecting mathematics with engineering and physics:
"Assume you have a SQUID placed close to a sample, placed inside a Helmholtz-coil pair. These coils are to be controlled by a second SQUID in a certain distance from the first squid, and far enough from the sample not to be affected. You observe that the signal on the first one fluctuates erroneously, without the signal in the second quid fluctuating. Your task is to identify the problem. Please tell which calculations you would to first and which experimental and construction data you would ask for."