Traditional psychological assessment has reduced the hiring and promotional error rate in modern businesses successfully for decades. They have also been used extensively to identify ‘hidden talent’ or ‘potential’ in people with limited work experience such as graduates, and also applied as a means for identifying future leaders at different levels of seniority, as well as in succession planning.
Psych testing is essentially an old-school form of predictive analytics, but they are limited in insight, providing a test of your ability to do a test. That’s it. Traditional psychological assessments do not link to actual performance in the role, nor do they have any self-learning functionality. There is no performance data that feeds into psychological assessments and therefore they have limited predictive power and no learning capability.
The worst aspect of psych tests is that you need multiple tests to test for multiple attributes. This is because they are just not that smart. This is where innovation necessarily disrupts an old formula. The difference lies in the data – volume and variance. A psych test is usually multi-choice questions repeated in different ways to achieve validity. You and I might pick the same option for each question and the only way to distinguish between you and me is to ask us a lot of questions and hope we pick some that are different to recognise our differences.
Data that comes from free-text answers to open-ended questions is by definition going to be hugely varied. A question like ‘what’s a favourite experience of working in a team’ asks us to each delve into our own personal experience, a behavioural interview question which means our answers will naturally be different.
This formula of using data that is uniquely personalised delivers variance that psych tests just can’t deliver. Ever. When it comes to developing an Ai based assessment the questions that a candidate is asked, and the answers to the questions are suitably diverse, psychologically robust and designed with the same rigour in standardised Psychological assessments.
With the processing power and advances in Natural Language Processing (natural language being the origin of all psych tests) instead of having to force a candidate through multiple tests you can distil many attributes from one test. That test is usually 20 minutes, asks 5 questions, with up to 80 features able to be discovered about that candidate including their critical thinking, their drive, self-awareness, accountability and team orientation, their propensity to stay in a role or not, their HEAXCO traits and their communication skills.
The ability to better understand individuals based on their answers to questions means we can provide accurate and insightful feedback to everyone within a couple of hours. Feedback allows everyone the opportunity to be heard, understood and cared for. This is equity.
It’s a cliché, but nonetheless true, that as time passes all processes become dated.
Some might need to be thrown out completely. Many more need to be adjusted and refined to keep up as workplaces and ways of working change.
I’m not old enough to remember the recruitment days of Rolodex and faxed documents. But I’ve heard the stories. Paper mountains of resumes teetering on desks. Consultants queuing at the one office fax machine to send their applicants’ profiles to clients.
Who knew that today we’d be communicating almost instantly by email, on our own computers, or sifting through resumes using Applicant Tracking Systems? In the 1980s that would have sounded like something from Doctor Who.
Since then, it’s all slowed down a bit.
Sure, ATSs take a lot of the legwork out of choosing who to interview. But they’ve also led to Resume Optimisation tools to help applicants beat our filters.
How can we avoid picking only the people who are best at gaming the system? How do we know we’re not missing our perfect applicants?
Now AI is taking the hiring process another leap forward. It’s speeding up the more process-driven elements and helping us select interviewees who are more likely to fit into our businesses.
And that means we need to re-examine two elements of that hiring process – the resume and the interview.
First, let’s tackle the resume.
Here’s a challenge for you. Find five well-known businesses that don’t ask for a resume on their careers page. Difficult, isn’t it?
Now think about the resumes you’ve seen recently.
I’ve seen resumes that are well-constructed, professionally crafted prose. And others that are complete works of fiction.
You’re as likely to find glaring spelling mistakes, a messy layout, and a shameless plea to be considered as you are a concise summary, an attractive photo and carefully chosen keywords. If you’re really unlucky you get all of these in one “super-resume”.
A quick search on “How are resumes used?” reveals the astounding advice that applicants should “know the facts in detail, as they may be questioned” about them. That just confirms my suspicion that these documents are more like scripts than records of facts.
And, there’s one more thing that recruiters know about resumes, even if they don’t all admit it …
According to research by the Cambridge Network, some recruiters give CVs a six-second speed-read and many recruiters spend just under 20% of their time on a profile … looking at the picture!
Resumes are rarely used correctly or understood properly, by applicants or recruiters. They most certainly do not predict how successful an applicant is likely to be in a role. Instead, they’re a minefield of potential bias: year of graduation (age bias), name (racial / gender/identity bias), experience in a similar business (confirmation bias), and so on.
So isn’t it better to put some truly intelligent AI for HR to work instead?
I was astonished to see that 96 per cent of senior HR leaders understand the benefits of using artificial intelligence in their HR and talent functions. But there’s a big gap between recognising the benefits and reaping them.
The canny HR leaders who are already adopting AI techniques will have a head start on their slower rivals.
Some more traditional HR tech providers have evolved their recruitment tools, presenting them as predictive. However, they’re more likely to be creating profiles of your better staff and matching these profiles to the external candidate market, not predicting how they will perform.
Instead, the new wave of HR tech uses well-constructed algorithms, created using a business’s performance data, to provide an unbiased shortlist of candidates far more likely to succeed within the business once hired.
The algorithm can’t be misled by optimisation techniques, personal feelings or prejudice. Instead, it uses objective data, science and evidence to find the people who are most likely to be a good fit and perform. For this role, in this business. And it will help uncover applicants we might have otherwise overlooked when their resume didn’t match our expectations.
The better solutions work by identifying the defining characteristics of the whole performance group within a business (superstars through to under-performers) and then predicts where external applicants will sit on your performance scale once/if hired.
These advanced solutions then go further via validation reports to prove their better predictions are turning into better new hires. They then use Machine Learning to ensure each unique model continues to learn more about the performance of each business, further improving its predictive power over time.
These two additional steps mean that whilst us humans are still required to make the final hiring decision, we will get better results for our applicants and our businesses. Maybe that’s where the resume might still have a role – as the frame for some reasonable high-level questions to help us understand the person in front of us in more depth, once they’ve got through the first stage.
The most sophisticated algorithms are already outperforming humans in the selection and identification of suitable candidates – and by that I mean candidates who go on to become productive, valuable and loyal employees.
So, what would you rather have?
– A shortlist of candidates chosen because of what they’ve selected to include in (and omit from) their resume?
– A shortlist of candidates you know are likely to do well in your workforce, because they’ve been chosen using statistically-proven, company-specific performance drivers validated by behavioural science?
Not that tricky a question, is it?
And very easy to see how, with the advent of AI for HR, resumes will soon be as much a part of recruitment as faxes and Rolodex.
The Americans with Disabilities Act (ADA) passed in 1990. This year, Australia’s Disability Discrimination Act turned 30. Even after all that time, bias and discrimination against candidates and employees with disabilities continues to be an important topic.
The unemployment rate for those with a disability (10.1%) in 2021 was about twice as high as the rate for those without a disability (5.1%) (U.S. Bureau of Labor Statistics, 2022). Coupled with increased laws and regulations regarding the protection of disabled job applicants and employees (e.g., U.S. EEOC, 2022), it is no surprise that academics, employers, and selection vendors are keen to understand where potential disability bias exists so it can be reduced or, ideally, eliminated.
Traditional face-to-face or video interviews in particular create potential barriers for individuals with disabilities, due to the well-documented stigma and prejudice against those with disabilities (Scior, 2011; Thompson et al., 2011). One study found that fake accountant job applicants that had disclosed a disability were 26% less likely to receive employment interest from the employer than those with no disability. Worse, experienced candidates with disabilities were 34% less likely to receive interest, despite presenting equally high levels of qualifications (Ameri et al., 2015). In addition to the bias held by hiring managers or recruiters, another concern is that certain selection methods create a very poor candidate experience for individuals with disabilities, causing them stress or anxiety and therefore stopping them from putting their best foot forward. For individuals with Autism Spectrum Disorder (ASD) in particular, in-person or video interviews can be very stressful, with less than 10% believing they are given the opportunity to demonstrate their skills and abilities in this process (Cooper & Kennady, 2021).
Stuttering is another form of disability where traditional in-person and video interviews where the candidate has to speak may lead to stress and anxiety (Manning and Beck, 2013). One study found that people who stutter find their stuttering to be a “major handicap” in their working lives and over 70% thought that they had a decreased opportunity to be hired and promoted (Klein & Hood, 2004). Other disabilities, such as dyslexia and other learning and language disabilities may cause candidates to struggle with timed online selection assessments, so it is important to identify and remove these barriers (Hyland & Rutigliano, 2013).
Cooper and Mujtaba (2022) recommend alternative approaches that allow candidates with ASD to showcase their skills without having to verbally communicate them or properly interpret nonverbal cues.
The use of an online, untimed, chat-based interview – that is, our Ai Smart Interviewer – can not only help reduce discrimination against those with disabilities but also create a more positive candidate experience for them.
This format is particularly helpful for individuals with disabilities where traditional in-person interviews, video interviews, or timed assessments may cause stress or discomfort, therefore not allowing the candidate to express themselves freely and adequately demonstrate their skills.
Our Sapia Labs data science team has submitted a paper on reducing bias for people with disabilities to SIOP for 2023.
In the study, the data comes directly from our Smart Interviewer, which, as we said above, is an online untimed chat-based interview platform.
Candidates can give feedback after the interview process, and some candidates include self-report disability conditions in their feedback. While a number of different disabilities were mentioned, we had sufficient sample sizes to examine candidates with autism, dyslexia, and stutter. We compared their machine learning-generated final interview scores and yes/maybe/no hiring recommendations to a randomly sampled, demographically similar group of candidates that did not disclose a disability.
Effect sizes, 4/5ths ratios, and Z-tests revealed no adverse impact against candidates with autism, stutter or dyslexia. Additionally, feedback from these groups tended to indicate the experience was positive and allowed them the opportunity to do their best.
True diversity and inclusion starts with the way you hire. Our Ai Smart Interviewer allows people with disabilities and neurodiversity – real people, with real ambitions – to represent themselves fairly.
A job interview is often an intimidating experience for a candidate, but it needn’t be this way. There are ways that companies can make interviews a comfortable process for the candidate, more effective at getting the right data to make decisions, and reduce biases that can disadvantage members of under-represented groups.
Interviews need to be structured and ask the same standard questions of everyone, making them applicable to the type of role you’re filling. Questions need to be open-ended that permit more than one answer, providing an opportunity to see how candidates think through problems and solutions. Questions shouldn’t be written to be as “gotchas,” but rather give people an opportunity to be themselves.
We’ve talked at length about bias when doing initial screening, but this is something that traditional style face-to-face first interviews also don’t solve for. It is possible for interviews to be ‘blind’ and free from bias as well.
This requires removing visual biases – those based on what we see – from an interview process. This is made possible through the use of text-chat as the preferred method of interview. It’s something that many successful companies like Automattic (the makers of WordPress) have done for years.
Texting is something that most people are familiar with. Ai-enabled text chat feels very similar to texting a friend. Text chat is how we truly communicate asynchronously – we all do it every day with our friends and family. It needs no acting; we all know how to chat. Empowered by the right AI, text chat can be human and real, it is blind, reduces bias, evens the playing field by giving everyone a fair go and gives them all personalised feedback at scale. It can harness the true power of language to understand the candidate’s personality, language skills, critical thinking and much more.
We know we can get this right because at Sapia where we use chat-enabled Ai we send every candidate who uses our platform feedback on their interview, identify their strengths and weaknesses and help them understand what they might improve on. Thousands of messages a day confirm we are accurate (98%) of the time.
An inclusive interview process doesn’t exclude anyone from having an interview. This is something we are able to offer at Sapia. Everybody gets a chance at interviewing for the job. Everyone gets a fair go.