[website] BBC Viewpoint: More women needed in technology

[pcstru]

By interviewing them? It becomes very obvious in an interview whether someone would be good or bad at a job. It takes careful design of questions though.

 

I once got into terrible trouble when defining a recruitment process. We had met to decide questions to ask the candidates and I suggested sending the questions along with the job descriptions. I said that would give each candidate as much time as possible to think of good answers, so we would be judging the best quality response the candidate could come up with rather than conflating the best response in a rather short interview time during which the candidates might be stressed. That was met with some hostility but they could just about see my reasoning. Unfortunately I completely lost them when I went on to suggest that we also send candidates our model answers. That went down like a lead balloon and I doubt they could see the point even with a large telescope.

 

Interviews are terribly difficult to do well and are easily subverted by unconscious bias. They are usually the least rigorous part of a recruitment process and interviewers are often unskilled and don't even attempt to differentiate between someone parroting what they expect to hear and someone who genuinely knows what they are talking about. Recruitment metrics are often poor but without good measurement of outcomes, it is not obvious that good metrics are any better than poor ones.

 

There is good research on bias in recruitment selection using blind controls to show how prevalent unconscious bias on gender, race etc is. There is also plenty of evidence in the demographics of organisations that bias is alive and well. It's not just about women or particular ethnicities choosing not to do those things, it is (often unconscious) discrimination. If not positive discrimination as a means to address those imbalances then what?

[witch]

There is good research on bias in recruitment selection using blind controls to show how prevalent unconscious bias on gender, race etc is. There is also plenty of evidence in the demographics of organisations that bias is alive and well. It's not just about women or particular ethnicities choosing not to do those things, it is (often unconscious) discrimination. If not positive discrimination as a means to address those imbalances then what?

 

I have done a great deal of training on recruitment and I agree, there are many examples of unconscious bias in the process. This is usually addressed, so far as is possible, by having more than one person in the process so that bias is counteracted. Bias is however a fact of life.

Unless you are sure that the discrimination is, for example, racial, how can positive discrimination help?

The answer is proper training - when I started in recruitment I had no idea how easy it is to be biased - my training was very good and faced me with countless examples of unconscious bias and how to counteract it. You can't be careful of it if you dont know what it is or what to look for.

[Oaktech]

and another gem from the BBC... BBC News - Ada Lovelace: Challenging women's status in science

[pcstru]

@witch, Again I agree (at the risk of ruining my destructively argumentative reputation!) , Quality training should go a long way to addressing bias but how many people actually do recruitment training? Also if you are faced with kind of bias in (say) the London Met, which was branded "Institutionally Racist", do you really think training would address the issue?

 

I'd also like to be clear that I'm not arguing that positive discrimination should come before merit. I'm suggesting that merit can be specified and that you can end up with many candidates who are equally good, equally capable. When you cross that line, discrimination on any basis is fairly arbitrary, so why not allow a demographic imbalance to be addressed.

[witch]

@witch, Again I agree (at the risk of ruining my destructively argumentative reputation!) , Quality training should go a long way to addressing bias but how many people actually do recruitment training? Also if you are faced with kind of bias in (say) the London Met, which was branded "Institutionally Racist", do you really think training would address the issue?

YES - if it was done properly!! THAT is the problem, and THAT is what needs to be addressed. Many people don't realise their unconscious bias until faced with it and that alone will help immensely. It also flags up the people who are possibly going to be a problem where recruitment is concerned.

After all, who is deciding what level of positive discrimination should be applied? Surely the accusation of bias would apply to them as well?

[localzuk]

I once got into terrible trouble when defining a recruitment process. We had met to decide questions to ask the candidates and I suggested sending the questions along with the job descriptions. I said that would give each candidate as much time as possible to think of good answers, so we would be judging the best quality response the candidate could come up with rather than conflating the best response in a rather short interview time during which the candidates might be stressed. That was met with some hostility but they could just about see my reasoning. Unfortunately I completely lost them when I went on to suggest that we also send candidates our model answers. That went down like a lead balloon and I doubt they could see the point even with a large telescope.

 

I'd disagree with both of those things. I'd not want to send candidates the questions we're going to ask them. I'd send other, different, questions. I'd also not understand sending model answers. As that just means someone could be a good liar and copy them in a clever way, same with the questions - send the ones in advance and they could just be good actors, repeating things they've been coached to say.

[pcstru]

YES - if it was done properly!! THAT is the problem, and THAT is what needs to be addressed. Many people don't realise their unconscious bias until faced with it and that alone will help immensely. It also flags up the people who are possibly going to be a problem where recruitment is concerned.

It possibly does flag them up, but in (say) the London Met, the people they were being flagged up to were simply dismissive of the issue - "bias, what bias?". "institutionally racist" meant that racist attitudes were endemic from the top of the organisation to the bottom. It wasn't just overt racism that was a problem, the dominant culture was of white police officers who's most common experience of "black culture" was in street level confrontations. I'd like to think that training could address the issue but I think at best it would be rather slow and at worst it would have had very little effect.

 

My experience of private companies is that recruitment training is rare. Even in the public sector the requirement is that one interviewer must have done fair recruitment training, it's not required for all interviewers to have done it.

After all, who is deciding what level of positive discrimination should be applied? Surely the accusation of bias would apply to them as well?

 

Sure, and there is a huge irony in discriminating on the basis of X to address imbalance caused by discrimination against X. The hope is that once the imbalance in an organisation is addressed, that itself causes a culture shift and balance is maintained without the need to positively discriminate on X. I suspect it's not quite that simple in practice.

[pcstru]

I'd disagree with both of those things. I'd not want to send candidates the questions we're going to ask them. I'd send other, different, questions. I'd also not understand sending model answers. As that just means someone could be a good liar and copy them in a clever way, same with the questions - send the ones in advance and they could just be good actors, repeating things they've been coached to say.

 

The suggestion had two purposes. First to get the interviewers to think about a situation where they are interviewing someone who is giving them exactly the answers they expect. We were looking for Prince 2 skills for which there were plenty of courses (so plenty of opportunity to learn to talk the talk) and Prince 2 is very particular about language such that you are steered to home in on particular keywords. Second and more important was to try and get them thinking about the list of questions and the whole idea of 'model answers'. The fact that we had model answers to many of our questions was an indication of a poor interview process. You have model answers for an exam or test, when you are interviewing someone you should be trying to explore their experience and your questions should such that you are unlikely to have model answers. A reasonable interview question might be along the lines of "tell me about your strengths and weaknesses" or "can you tell me about an experience you have of handling a challenging situation". They invite the candidate to relate actual experience and there are seldom right or wrong 'model' answers.

[localzuk]

The suggestion had two purposes. First to get the interviewers to think about a situation where they are interviewing someone who is giving them exactly the answers they expect. We were looking for Prince 2 skills for which there were plenty of courses (so plenty of opportunity to learn to talk the talk) and Prince 2 is very particular about language such that you are steered to home in on particular keywords. Second and more important was to try and get them thinking about the list of questions and the whole idea of 'model answers'. The fact that we had model answers to many of our questions was an indication of a poor interview process. You have model answers for an exam or test, when you are interviewing someone you should be trying to explore their experience and your questions should such that you are unlikely to have model answers. A reasonable interview question might be along the lines of "tell me about your strengths and weaknesses" or "can you tell me about an experience you have of handling a challenging situation". They invite the candidate to relate actual experience and there are seldom right or wrong 'model' answers.

 

Asking questions like those last 2 is great, but it is not going to allow you to figure out if someone is suitable for a technical position. You need to ask them about experiences of specific technologies, or specific situations. Especially as those 2 can basically be lied about.

 

Eg. I asked questions in an interview a few months ago such as "You are asked to filter a website for a specific room. That room does not have specific PCs assigned to it. Instead, machines are wheeled in and plugged in in the room. You can also not filter by user, PC or IP as these will all change. How would you fulfil this request."

 

I had an idea, as I'd already done it myself (the ports in the room were put on their own VLAN, and that VLAN's IP range was then filtered), but I wanted to see how others would approach it. It highlighted knowledge of networking technologies, and knowledge of filtering techniques.

 

Hence there being a model answer...

[pcstru]

@localzuk : That's a fair point although I personally would look to present the kind of questions which do have model answers in an exam/test prior to interview. I might then use the answers as the basis for some discussion in the interview but in the main, asking those type of questions in an interview seems to me to be (possibly) poor use of the time.

[localzuk]

@localzuk : That's a fair point although I personally would look to present the kind of questions which do have model answers in an exam/test prior to interview. I might then use the answers as the basis for some discussion in the interview but in the main, asking those type of questions in an interview seems to me to be (possibly) poor use of the time.

 

Anyone can prep themselves to answer questions if they're given enough time. Part of the whole point of an interview is putting them on the spot. Seeing their deductive reasoning abilities, seeing if they are capable of thinking on their feet, or whether they just resort to Google for everything.

 

Its a good use of the time IMO.

[pcstru]

Anyone can prep themselves to answer questions if they're given enough time. Part of the whole point of an interview is putting them on the spot. Seeing their deductive reasoning abilities, seeing if they are capable of thinking on their feet, or whether they just resort to Google for everything.

 

I don't think we are in much disagreement. You will ask the question in interview and then probe around the answer to assure yourself that they do actually know what they are talking about. I prefer to do those questions pre interview and use the interview itself purely for the final assurance. Perhaps you do longer interviews so time isn't an issue.

[PiqueABoo]

Just catching up with this thread. Can't believe it got diverted into generic equality blather so quickly without anyone linking *THE* video.

Slackers ;b

 

http://www.youtube.com/watch?v=g032MPrSjFA

Edited October 18, 2012 by PiqueABoo