Adverse Impact and Hiring

What do we mean by Adverse Impact?

By Nick Jarman – Research Assistant at Cognisess

…Adverse Impact is most pertinent when it has a negative effect on candidates of a protected class, namely: Age, Disability, Gender, Race, Religion, Sexual Orientation, Marital Status, Pregnancy/Maternity.”

Any candidate, for any role, within any organisation, could potentially be adversely impacted by the way they present themselves at each stage of the selection process.

This Adverse Impact is most pertinent when it has a negative effect on candidates of a protected class, namely: Age, Disability, Gender, Race, Religion, Sexual Orientation, Marital Status, Pregnancy/Maternity.

Adverse Impact can be felt even during apparently neutral processes, a measure of indirect discrimination.

An important aspect of indirect discrimination is that it is often identified statistically. The intention to discriminate is irrelevant. The output of the selection process is the most important factor.

It is worth noting that a policy is not indirectly discriminatory if the action is objectively justified in the absence of a less disadvantageous method of producing the same measure.

Although Adverse Impact can be measured during each stage of a selection process, a general rule of thumb is that the hiring-rate of each minority group must be 80% of the hiring rate of the majority group within that company. For example, if males comprise the majority class within the organisation and 90 males are hired from a pool of 100 then the hiring rate from each other minority group must be at least 72%.

How can we mitigate Adverse Impact in our hiring process?

There are a few general ways in which Adverse Impact can be reduced during a selection process. For one, regular monitoring of pass rates at every stage of the selection process will allow apparently neutral policies to be statistically explored regularly and often. The result is a process within which any indirect discrimination can be detected and rectified before it causes Adverse Impact for any particular class of candidate.


Another method of reducing Adverse Impact in a selection process is by focusing specifically on job-relevant competencies. These might be identified as a result of rigorous job analysis based on the job description, or can be more generic competencies specified at the beginning of the hiring process. The result of a specific focus on job-relevant competencies is to judge every candidate on the exact same competencies, regardless of their personal characteristics, thus removing irrelevant stages.

Related to this is the use of standardised interview techniques. Rather than having a more free-flowing personal interview where the interviewer’s personality and individual interests may be a source of unconscious bias, the interview will contain set questions designed to measure aspects of the job-relevant competencies defined at the start of the hiring process. In addition, constant monitoring of any assessors used within the selection process will serve to identify any unconscious bias creeping into any decision making process before it becomes an Adverse Impact. It also informs the focus on ongoing assessor training.

A final way to help reduce Adverse Impact in a hiring process is to make use of a diverse panel of assessors. This serves a myriad of purposes, not least to help reduce any unconscious bias present within any individual assessor. Introducing a wider diversity of thought will also help reduce any naivety that might occur in establishing the selection process.

Request a demo to learn how Cognisess can address Adverse Impact in your hiring process. http://www.cognisess.com/request-demo

Contact Polly Hill to be kept up to date with Cognisess Insights. – polly@cognisess.com

The AI will see you now… why we should trust Computer Vision

On the whole, humans are generally good at identifying emotions, but this mostly happens at an unconscious level and it isn’t always a reliable tool, particularly when it may involve having to make decisions based on this interpretation. According to psychology professor, Lisa Feldman Barrett our emotions are the brain’s method of understanding the body’s raw data. But sometimes the same sensations can be interpreted differently depending on external events. For example, a queasy stomach could be interpreted as nervousness before a job interview or excitement for an upcoming holiday.

We are reactive to the world around us. For example, if a HR professional has a disagreement with a colleague or is experiencing personal issues outside the workplace, they may bring these negative emotions into interviewing a candidate for a job. This may affect our perception of a candidate and the hiring decision.

Although we can’t change these factors, we can prevent them from influencing the recruitment process to ensure it is fair, consistent and transparent. As we enter the fourth industrial revolution and are increasingly driven by new technology, AI and Computer Vision will be on hand to assist humans in making data-led decisions that are consistently accurate and free of bias; attributes which will become increasingly important in a fast moving and compliant world.

What is Computer Vision and AI?

Computer Vision acts as the eyes of an AI. This technology can visually process the world around them and they can be programmed to analyse the information it collects. For example, Computer Vision is used as part of our video analysis tool. The technology analyses each candidate’s facial expressions and micro-expressions from their video interview. Micro-expressions are facial movements that happen – even before we are consciously aware of making them. This means Computer Vision is able to detect a candidate’s true emotion at that moment, not what they may like to present to an interview panel.

This tool can be used to recruit across all industries. However, it is especially useful in customer focused or sales roles where ‘reading people’ or presenting yourself positively or empathetically is required. Computer Vision can observe how easy a person is to talk to, and how approachable they will be to customers – even under pressure.

Why should we trust machines to assess human behaviour?

Together, AI and Computer Vision are able to accurately repeat the same process over and over again. Our video analysis tool has analysed over 1.4 billion facial expressions with a 97% accuracy rate. This is an invaluable tool for humans and can be used to gain advantage particularly in dealing with scale or distance. Computer Vision doesn’t experience a post-lunchtime dip,  which is when a person feels drowsy for a few hours whilst their body digests their food.  Equally the AI doesn’t have a ‘bad day’ through fatigue which would impair their overall cognitive performance. It can process thousands of job applications and identify the top candidates, regardless of what time of day it is. This enables a recruiter to stay focused on putting their attention and expertise on the applicants that have relevant skills and abilities for the role.

As AI and Computer Vision aren’t human, they have the capacity to be completely objective (read our blog on unconscious bias in humans here). Our Chief Scientific Officer, Dr Boris Altemeyer, revealed to Information Age that “AI doesn’t have to understand its own unconscious bias, because it has none. AI does not need diversity and inclusion training. It’s incapable of taking an instant dislike to someone, secretly wondering whether someone’s planning on starting a family, or hiring someone who’s pleasingly similar to them.” As there isn’t any human involvement in the analysis process it supports a less bias hiring decision. Our Computer Vision isn’t programmed to see race, age or gender – it is purely assessing what an applicant is non-verbally communicating. This becomes a vital tool when companies are pursuing policies of inclusion, diversity and fairness.

Making fast and accurate hiring decisions for every candidate – every time

Our clients using our video analysis tool will be able to give each applicant a fair chance, regardless of how large their applicant pool is.

Each candidate records themselves answering a number of key questions that have been set by the company. Computer Vision analyses each video interview frame by frame, searching for positivity and expressiveness, both of which are measures of genuine passion. ​The tool is a hundred times faster and accurate than any human assessment. As a result, a business is able to identify top talent from their entire applicant pool.

Working with Computer Vision supports hiring decisions, HR professionals are able to draw conclusions from their own expertise when finally ‘meeting’ each candidate, but also review the Computer Vision comprehensive analysis on them alongside the other Cognisess assessments that analyse Job Fit, Culture Fit and Team Fit.  

Recently, we have studied the performance of the video analysis tool with our client, AB InBev – the world’s leading brewer. Focusing on the candidates who had successfully passed their video interview stage, we uncovered a statistically significant relationship between an internal assessor’s high rating of a candidate and a good positivity and expressiveness score. This further shows how accurate this technology can be and how it can be applied to create a fair and transparent recruitment process were any areas of unconscious bias can be corrected as a helpful correction mechanism to support HR professionals avoid bias.

Make every day a good decision day with AI

Everyone of us within a workforce will have good and bad days, it is an inherent part of being human, after all. But in this new age of technology, AI and Computer Vision will ensure those occasion bad days doesn’t turn into bad hiring and bad decision days. If you would like to find out more about how our video analysis tool or Computer Vision can transform your recruitment process, book a FREE demo today with our expert team.    

3 Key Questions to Consider Before Beginning Your People Analytics Journey

Cognisess’ Chief Scientific Officer, Dr Boris Altemeyer answers three key questions addressing what companies who are thinking about embracing People Analytics need to consider. The insight was based on our recent case study project with hotel giant IHG.

Why should companies put their faith in People Analytics?

For decades the HR hiring manager was the best people analytics capability companies had. They worked without any data backup, operated based on a gut feeling and what had come before. Gut feeling is brilliant but it is the imperfect and human equivalent of processing lots of data and coming to a conclusion. Except….if that hiring manager has a bad day it will affect who’s going to get hired and if they leave their job that data model leaves the company with them.


Cognisess’ Chief Scientific Officer, Dr Boris Altemeyer

As a Talent Management platform, Cognisess brings a systematic approach which can learn and analyse a company’s people preferences and understand what truly makes someone great in a job role, in a team or in that culture – which is what hiring managers are trying to internalise.  Once the machine has learnt, it is able to be totally objective and it doesn’t have a bad day! If the company changes direction or focus – it doesn’t matter, the AI is much more flexible and adaptable than humans will ever be. The beauty of Cognisess is that we can take the best of what makes us humans human and equip HR to make better decisions backed with data and science. That makes for a really exciting and fulfilling future for HR, they will be able to better demonstrate and validate the value of its human capital for the whole organisation and key decision makers.

Our client IHG was really receptive to using People Analytics. They wanted to source the right people who would predictably stay with them, develop in their roles and move them forward within the organisation where their talents were best suited. But there was also a time and resource pressure which left little room for error or waste. Cognisess Pro was able to provide for these competing pressures.

Prior to using People Analytics, what do companies need to do to prepare for it?

Be open-minded. The data companies have may be valuable to the process – or they may need to set up a new method to collect and collate that data over time. People Analytics isn’t a silver bullet that businesses can implement instantly and become 50% more efficient – it’s an ongoing, organic journey. Using data to inform your decision making is not a quick-fix, but about adopting a new mindset. The technology opens up a lot of questions and insights, which can be interesting and inspiring, they are part of kick-starting the process and draw the business’ attention towards just how powerful People Analytics can be.

This was something we saw in IHG’s Future Leaders programme. From the beginning of the project there was a world of information that was already available, but not as actionable data. This was absolutely invaluable for us. The data covered previous cohorts of candidates who not only passed the assessment centre, but were now operating in management positions. The assessment centre results from these brilliant high performers were used to benchmark what ‘good’ looks like. Additionally, we analysed the people who received really good reviews from their supervisors and had made career progress within IHG. This is usually dormant or inactive data that sits somewhere in an HR system but no one looks at – but this is the kind of information that the machine loves! People Analytics recognises that one person was promoted quicker than another and therefore they must be demonstrating some qualities which make them better.

Cognisess Pro’s clients are from a wide range of sectors. Does each sector present different needs for People Analytics?  

We learn an incredible amount from all our clients, regardless of which sector. In fact, it helps to have data sets constructed across many sectors and industries. A university professor half jokingly said we are becoming proxy experts in so many industries. For instance, we have gained an understanding of what best looks like in the hospitality industry and the automotive sector – but it doesn’t mean we intend to run a hotel or build a car. Our focus is very much on understanding people, behaviour and potential. The more understanding we have, the better we can design the system with the widest user experience at the heart of it. We are also very focused on making very complex data accessible, which helps clients make the very best people and talent decisions for them.

If we look at very different clients like IHG and Vauxhall, their applicant journeys are very divergent. IHG’s candidates will fine tune their applications and apply to a select few graduate programmes – as there’s not many around. Vauxhall applicants are more likely to consider hundreds of sector openings. They aren’t likely to complete a 35-minute application when competitors only want them to answer a few questions. Vauxhall’s user journey is a lot quicker than IHG’s. With IHG we designed a more ‘Intercontinental Hotels’ experience for the candidate, as they want to spend more time on their application.

We aim to have a successful and frictionless journey for all applicants whilst ensuring we get the right amount of data for each client. You can only do that if you can draw your experience from as many sectors as possible, we look forward to adding many more sectors to build our reach and understanding. I am confident that whichever sector we serve we will be able to grasp the minute variants and dynamics that are at play there, we will help deliver People Analytics as a driver for transformation and performance for those clients.

For more insight, download our recent case study with global hotel giant IHG.