@eden put the men resumes directly into the trash then strip the names off.

@eden Regarding underestimation and overestimation of skills, that information in some cases may be worthless. Try to find out whether the experience they claim to have was actually applied. There should be an inconsistency if they clame to be helicopter technicians but the only jobs they've had in the last five years were as a chef. I mean, not such a stark contrast, but the inconsistencies help to find the truth.

@h This was one of the suggestions I quite liked, and I'm going to spend Monday trying to remove Names from the system we have so that I just get an ID.

With your reply to under/over estimate, this is when I just have a CV and I'm trying to filter for who's worth following up with an interview.

It's easy to do consistency in the interview, we have good tests for that, but beforehand I want to put people through fairly.

@eden Glad that helped then 😃

@h This would mean just widening the criteria for who goes to face2face interview.

I think that's the only solution? Allow wider requirements for diverse groups to get them to stage 2, and see how they hold up in an assessment.

@eden Maybe you need three stages, make many more shorter ones over the phone as to not impose an excessive cost with too many in-person interviews.

@h So we have two phone interviews and an in-person right now:

* Intentions & Fit - let them understand the company and see if they're appropriate
* Team Fit - Talk to a department lead & see how they come across
* Technical Fit - In person interview.

So there shouldn't be anything too bad, but it's hard to know if their CV isn't congruent with ability until the third bit!

@eden I don't know if it's possible in your line of business, but in my case for example, in the software industry we can test technical ability remotely, and in fact we usually do.

I reckon that's not something you can do in every industry, so that would probably have to be something tailored to the limitations of the type of work you're hiring them for.
(If it's writing, accounting, or other type of knowledge work, that's testable remotely. It's harder for physical work for obvious reasons) )

@h We could. This is Data Science, so it's useful to know if someone can write suitable code, but we find it a lot more helpful when we get to go through section by section and work out how much they understand of what they're doing compared to how much is plug n play.

But it's definitely possible, and I think it'll be worth adding in quick little Python/R tests :)

@eden Then a preliminar filter of technical ability at the very beginning should be helpful to screen people who wouldn't make the cut with the more complex stuff that's necessary to get the job done.

As a bonus, you may also be able to see other things in their work early on, like creativity, resourcefulness, attention to detail, etc.

@eden At the same time, you don't want internet access and access to a powerful computer to be a barrier to entry.
People who can't afford a powerful computer, or people who live in areas badly served by internet providers would be at disadvantage.
Remote tests are tool to be used with good judgement, as any other.

@h Yup. We basically ignore whether someone trained a model or not, and go purely on whether they showed capability and understanding of how to train one.

Definitely going to try out the mini technical test though :)

@eden Glad that helped. Cheers 😃

@Efi Regardless of how I feel about this, I like having a job, gender is a protected characteristic in UK law.

@eden aw D:

@mikeburns Oh. Like we look at clusters?

So we wait until we have a minimum of 5 CVs from women, and then take 10 CVs from men, and only consider that cluster?

I actually really like that?

@eden
Yeah! (Except 10 from women for every 5 from men ...)

I've heard rumor that the company Glitch does this.

@mikeburns The sad truth is that if we tried that, we'd never manage to do any hiring...

@eden @mikeburns
You can easily run into corner cases that way. Say for example John presents himself as a white man. He doesn't have to disclose he's a Trans or queer person. Clustering John together with the majority would hardly give any justice to proportional representation.

Unless I'm misunderstanding something, which is quite possible anyway?

@h @mikeburns it's unlikely to even be a corner case. Someone I work with right now identifies as NB but uses he pronouns and has a male sounding name, so it's just something we're going to miss.

If we did have that information, then we're can only be as fair as the information provided. If someone doesn't disclose we can't treat them any differently..?