Saturday, September 22, 2012

When the Hiring Boss Is an Algorithm

I can't claim to be an industrial psychologist, but I have worked with them and written much of the software that does some of the predictive analysis that was mentioned. I can claim to have a lot of experience with how these tests work.

I have to disagree with you right off the bat here: Employment screening tests are not a sham, and there are many good things about them.

- Has value, but primarily to the employer

Some of the predictive factors that show immediate value are used to best utilize an employee: whether they prefer working solo or in a group, whether money or recognition motivates them, if they are highly detail oriented, or whether or not they'll rock the boat (which could be a good thing! - like suggesting a more efficient process to replace an old, tried, and true one). This helps a manager provide a working environment that best allows an individual to excel, or to position people in groups so any perceived faults are covered with an overlap.

What also ends up being really important in these tests - to employers at least - is whether or not these people tend to lie, steal, or cheat, to abuse drugs or alchol, or may simply be reliable or not. Yes, if you crunch the numbers, you can take a good guess and produce a weighted prediction about this just from a personality profile.

When you're hiring for Walmart or Home Depot or some other vast chain with a large population of unskilled workers, weeding out the likely-to-be-bad ones shows a real financial impact, tens or hundreds of millions of dollars a year, statistically proven. Though the damages may not reach into the millions, smaller business owners are even more impacted by this.

- Must be applied properly

Some businesses, like Keller Williams Reality, do not apply their personality profiling properly. They use a Jungian based personality matrix - simplified slightly - to pigeonhole an applicant. If you do not fit the hole specified for the job, you don't get the job. They even make a big deal about how everyone applying for every position is required to do this, even if it's the next CEO (which was probably a lie, but c'mon, these folks are salepeople at heart). However, they decide in advance which single personality a given job requires, and if you don't match it exactly, you're out. So if you apply for a programmer position, and you're an extrovert, you won't get the job. If you're creative - you don't get the job. You have to correspond to THEIR sterotype.

Obviously this is wrong.

As the parent poster can probably tell you, the proper way to do it is to have your current employees take a test, sort by role, and attempt to find people who are close to the personality traits that your star individuals have in common. You'll also have to update this over time as market or work environments change.

This works because we're producing sample data, isolating trends, and using it to predict success based on commonalities. It's simple statistics. This is how we can 'catch' drug abusers and thieves before the fact: using the personality profiling tests data from criminals, we can find people who match their common traits. It may sound harsh, but the false positive outliers are exactly that - outliers. The vast majority is predictable to a reasonable degree.

Application of this information is a standard practice in risk analysis. If you're hiring for a casino dealer, and your applicant shows up as being 80-90 out of 100 match with career criminals, maybe consider a bit more carefully or do a full background check, or maybe just don't hire them - find a less risky applicant.

- Can be gamed

Most of these tests are straightforward. There's rarely any tricks or clever 'gotchas'. This isn't like a police interrogation where they're trying to trap you. They're personality profiles, and they don't have 'correct answers'.

However, if you can successfully role take, you can determine the outcome without much guile. This is hard for many people to do, grante

Source: http://rss.slashdot.org/~r/Slashdot/slashdotScience/~3/m_ul2dpBw2I/when-the-hiring-boss-is-an-algorithm

mayan calendar nfl playoff picture nfl playoff picture rose bowl 2012 sat cheating scandal hangover cure lebron james engaged

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.