The "Un-testable Hypothesis" is that LGBT are more gifted at "reading" people than heterosexuals.
The hypothesis has its origins in the work by Ervine Gottman who speculated that stigmatized people(s) have to exercise care in announcing themselves to the world. Stigmatized people(s) must exercise discernment when announcing. For instance, pedophiles usually don't put bumper stickers on their vehicles that indiscriminately announce their sexual preferences to the world.
That speculation was then combined with the belief that "anything you do repeatedly you become better at" and the Un-testable Hypothesis was launched.
Since LGBT are over-represented in Human Resources departments, and since the Hire/Not-hire decision is critical to the success of EVERY organization, there is utility in choosing the "job interview" as the test-bed for evaluating the "Un-testable Hypothesis".
One group will be doing the interviews. They will be offered two scripts and will be told that the purpose of the interview is to evaluate the scripts. After each interview the members of this group will rate the candidates on a variety of metrics important to businesses like:
- Will be punctual
- Will get along with other workers
- Will not blow-off work days
- Will not steal
- Will not take short-cuts on defined work processes
as well as the interviewer's "certainty" of their assessment.
The second group will provide the people being interviewed. Before the interview, each candidate will be alone in a waiting room.
The picture in my head is that there will be some small, easy-to-conceal item in the room that many people would be tempted to steal. Say a twenty dollar bill or a cell-phone. Something desirable enough that 35%-to-65% will take it.
It may be difficult to get enough people to steal the item given the pervasive nature of security cameras. However, if the interviewee was offered $20 for volunteering and the bait was a $20 bill, that gives them them the cover story that "I thought that was how I was being paid" and might increase the theft rate into the desired range.
The metric of interest will be "Did LGBT interviewers identify the theft risk candidates any better than the non-LGBT interviewer?" with theft-risk being identified as the candidates who actually slipped the $20 or cell-phone into their pocket.
Since LGBT are only 5% of the US base population, it is desirable to enrich the sample of interviewers with additional LGBT members. That might be arranged by looking on LGBT sites like Grindr.
A reasonable alternative hypothesis to the Un-testable Hypothesis is that LGBT use a set of cues that are not consciously accessed by them. In order to exercise the possibility of that potential mechanism, it is desirable to ensure that the pool of candidates being "interviewed" are a mix of young, old, men, women, straight, LGBT, black, white, Hispanic, rural and urban, affluent, less-affluent, tats and non-tats.
It might be possible to gain insight into other biases or cleavage planes that guide LGBT personnel decisions.