Let’s face it; everyone sucks at screening resumes. One of the reasons why people are so terrible at resume screening is bias. This article explains why resume screening is useless.
The reason that resume screening is useless, is that most information on a typical resume can trigger all sorts of cognitive biases. These include:
We tend to like people who are similar to us in some way. You may think more positively about a candidate that has a similar background to yours. Maybe they went to the same school as you, or perhaps you share an interest. This can affect your decision to invite them to an interview, regardless of their qualifications.
We often interpret information that confirms our pre-existing beliefs, or specifically look for it, and ignore things that go against our preconceptions. If you are impressed that a candidate has a degree from a prestigious university with a great reputation, you are more likely to overlook the fact that they were not a good student and were barely able to graduate.
We can sometimes anchor on to specific things, and compare everything else to it. If the previous employee whose job position is now open was excellent, you can fall into the trap of comparing new candidates to the old employee’s achievements, and not judging the candidates on their own merits.
Fundamental attribution error
We tend to focus on personal characteristics and ignore external factors when judging the behaviour of others. If someone took a few years longer to finish their education, you might first think that they’re lazy, or stupid, but you may find out that they were fighting cancer at the time.
These are examples of just a few common biases that can happen during resume screening, or even later on in the hiring process.
Resume screening is useless. You might think that you can be objective and overcome your bias if you are aware of it. You can’t. Biases are unconscious and involuntary. If you think you can control them, then try not to think of a polar bear. Did you imagine a polar bear?
This psychological phenomenon is called ironic rebound and demonstrates how conscious attempts to suppress certain thoughts make them more likely to surface. It’s human nature.
So if the problem is human nature, perhaps AI is the solution? Bias is a human trait after all, so an artificial intelligence that screens resumes must be perfectly objective can’t possibly be biased, right? Wrong. In 2015, Amazon scrapped an AI recruiting tool that they were secretly working on because it was biased against women.
The computer models Amazon used were trained to screen for applicants using patterns in resumes collected over 10 years. Since the tech industry is mostly dominated by men, the vast majority of resumes were from men.
Amazon’s system learned to prioritise men and even penalised the keyword “women’s”, which would appear on resumes in cases of candidates who attended women’s colleges, or under their interests or achievements as “women’s chess club captain.”
While Amazon corrected this issue, they weren’t able to guarantee that their system wouldn’t find some other way to discriminate. Luckily, this system was never used. But, it turns out that even a machine learning algorithm will be biased if it’s trained using biased data.
Another reason why resume screening is so inaccurate is that most information on a resume is irrelevant. Almost all information on a typical resume has no bearing whatsoever on the quality of a candidate.
There has been plenty of research on the effectiveness of different hiring methods, and things like age, interests, education level, and job experience – details you’d find on most resumes – have almost no correlation at all with job performance.
This might be obvious for age and interests which aren’t directly related to most jobs, but how can education and experience not affect job performance?
Education may mean that someone has the formal knowledge required for the job, but that doesn’t mean everyone with the same education level will be equally good at a specific job. Every domain of knowledge has better and worse performers.
Think of the last school you attended. Would you say that all of your classmates were equally bright? Did they all have the same grades? Would they all be equally good at the same job?
Experience simply shows how long a candidate spent at a specific workplace, it doesn’t necessarily reflect how good they were at that particular job, or what they’ve learned. Think of other people who have the same job as you. Are they all equally good at what they do?
And of course, people lie. According to a ResumeLab study, 93% of people lie on their resumes. It’s easy to understand why – they’re trying to make themselves look as good as possible after all. One of the most common resumes writing tips is to tailor your resume to the job you’re applying to. A resume is not an objective overview of someone’s accomplishments – it’s a sales pitch.
An experiment conducted by Aline Lerner showed that neither recruiters, nor engineers, nor hiring managers can accurately assess the quality of candidates based on resumes. They were able to correctly filter candidates by resumes only about 50% of the time, which is as accurate as a coin toss or randomly throwing the “unlucky” half of resumes into the trash.
What can you do instead of using resumes?
Resumes are so entrenched as a standard part of the hiring process that it’s hard to imagine what things would be like if they didn’t exist, both as a candidate and as an employer.
How do you apply for a job if you don’t send your resume? How do you select candidates if you don’t have a summary of their careers and qualifications? What should you ask candidates to send you when you write a job advert?
An application form is a slightly better alternative, as you can standardise it so that you ask all candidates to provide the same information in the same format. But it’s still subject to the problems described above.
A better way is to start the entire hiring process with a short online test. There are 2 types of tests that, according to research, are the most effective hiring methods you can use: Work-sample tests, and General Mental Ability (GMA) tests.
To test if a candidate will be good at work — give them a sample of actual work to do. It’s a simple idea, but an incredibly effective one. All you need are a few short tasks that represent the kind of work a candidate would do if hired.
You don’t need to test for every skill in-depth, only the core skills required for the job. The test doesn’t need to be very detailed, a more in-depth test of skill can come later, as this is just the first step of the hiring process.
This way you will give candidates the chance to prove themselves right from the start, and eliminate those who are only good at promoting themselves without much to back it up with.
These tests measure cognitive ability, also known as the g-factor in Psychometrics, which is one of the most significant predictors of individual differences in education and employment. GMA tests are often referred to as aptitude tests or reasoning tests and can measure problem-solving ability, as well as abstract, logical, verbal, numerical, or other types of reasoning.
While cognitive ability doesn’t guarantee that a candidate has the necessary skills for the job, research shows that people with higher cognitive ability learn more job-related knowledge, and learn it faster, than people with lower cognitive ability.
Not only that, but the more complex the job is the more accurate GMA tests are at predicting job performance because major cognitive processes such as planning, judgment, and problem-solving are all used daily in many jobs.
You can also combine GMA and work-sample tasks into a single test if you want to save time. But keep in mind that, as the first step of the hiring process, the test should be relatively short as candidates don’t like taking too much time to just apply for a job.
To be more efficient you can use an online pre-employment testing service that automatically evaluates answers. There are plenty available, as automation of HR processes is increasing.
If you are sceptical about starting the hiring process with a test, consider the fact that this will eliminate the least suitable candidates immediately leaving you with a better-qualified pool of candidates to interview.
A short GMA or work-sample test doesn’t take too much time but gives you a lot of information about a candidate, so it should come first if you want your hiring process to be more effective and accurate.
Resume screening is useless; resumes are unreliable. You can’t trust the accuracy of their information, nor your own judgement when screening them. They’re not just useless, but often more counterproductive to your hiring efforts than you may realise.
Artificial intelligence can’t help either, at least not yet, since it tends to suffer from the same issues as the humans that created it.
Fortunately, it’s possible, and even beneficial, to avoid relying on resumes entirely.
Pre-employment tests provide more valuable information about candidates than resumes. Work-sample tests can evaluate a candidate’s basic skills directly, and GMA tests can show you which candidates will potentially be the best employees.
When used online, pre-employment tests can be used as an application process, and they can filter out lesser qualified candidates quickly, which leaves you with better candidates to interview.
About the Author
P.K. Maric works at TestDome, where he helps companies get the most out of their hiring process using automated skill testing. He also writes about hiring topics on TestDome’s blog, and occasionally other websites.