Using AI for Faster and More Accurate Assessments – Nova partners with Yobs

Posted on October 14, 2021

As I introduced in my previous post on Radar, at Nova we pride ourselves on having developed a technology (which we call Radar, following our galactic metaphor), that enables us to assess candidates at scale.

The Old Selection Process

Up until last week, the process of Radar that nominated candidates had to follow (yes, remember that Nova is by-invitation-only and that members have limited nominations) before getting into Nova included 5 key parts:

  • A Background section, which tries to understand the candidate’s professional and educational experience, as well as their international experience and their command of languages. Candidates are also asked to submit a motivational text explaining why they want to join the Nova Community. This takes candidates between 5 and 15 minutes.
  • A References section, which looks for external validation on the candidate. For this section, if you have a boss with a high professional profile vouching for you, that will give you more points than having a relative with a less relevant profile. This takes candidates between 1 and 3 minutes.
  • A Personality section, where candidates had to take a self-reported test on personality assessing the famous “Big 5”, the most researched and proved personality assessment model. This takes candidates about 3 minutes.
  • A Cognitive Ability (IQ) section, where candidates have to prove their logic and numeric reasoning skills in an inference-based test. This takes candidates between 5 and 10 minutes.
  • A Video interview, which assesses candidates based on 6 of the most demanded competencies: drive, leadership, problem-solving, adaptability, self-awareness, and communication. This takes candidates between 6 and 12 minutes.

 

What is changing?

Today, I am happy to announce our partnership with Yobs Inc. to replace our self-reported personality test with a professional psychologist-reviewed personality assessment based on the candidates’ videos, which are processed by Yobs’ AI.

The reason for this partnership is threefold:

  1. It will make the Nova assessment more accurate and therefore we will be closer to accomplishing our mission to find, empower, and connect the top talents of the world.
  2. It will reduce the number of steps on the Radar process, and make it faster (about 3 minutes less), thus reducing drop-out rates and the time candidates invest in the process.
  3. Finally, it will enable us to create a very insightful report for all candidates based on their personality, regardless if they are accepted or not into Nova.

Introduction to the Personality Assessment – The Big Five Personality Traits

At Nova, we have been using the “Big 5” model for years. This “Big 5” represents the five most important dimensions of personality and soft skills, following the Five-Factor Model (FFM). This model is used by 85% of the Fortune 500 companies in hiring and training assessments and is the most accepted by the industry and scientific community. It is also largely proven to be correlated with job success in many different occupations in consensus literature.

The Big 5 traits are:

  • Openness: Measures creativity, learning, innovation, and the ability to ride out change and turmoil.
  • Emotional Stability: Measures emotional reactiveness and vulnerability to stress. Neuroticism (the opposite of emotional stability) correlates with perceiving situations as threatening.
  • Agreeableness: Measures how well one gets along with others. It measures key behaviors like tolerance, friendliness, and cooperation.
  • Conscientiousness: Measures your attitude towards doing the right thing, for example doing a full detailed job even when the boss isn’t watching.
  • Extraversion: The tendency to feel most comfortable and seek the company of others, rather than solitude. This is the only trait analyzed that we don’t use in our scoring model yet.

 

A More Accurate Process: Self-reported vs. Psychologist-reported Personality

It might seem pretty obvious, but for years people have got this wrong in selection: candidates do not perceive themselves accurately, thus reducing the reliability of using a self-reported personality assessment. Research (Connolly, Kavanagh, & Viswesvaran, 2007) shows that self-reported personality assessments correlate (R~0.5) with informant-based reports, but there is a huge variance that makes those measures different when measuring on a case by case procedure as we do at Nova.

Discrepancies between self-reports and informant-reports can come from three possible sources: self-report biases, access to different information, and psychometric differences across informants. In general, candidates may have implicit theories that influence or their responses, which may bias their choice. Although access to different information makes sense as a potential reason (for example, the candidate might not be ok at the moment of the interview and simply did not show his or her true self), actually self-report biases are the most probable source of discrepancy for an assessment like the one we do at Nova.

Self-reported biases refer to the fact that candidates may incorporate social desirability in their responses, such that they present themselves as more consistent with social norms than their actual attitudes or behaviors. For instance, to the sentence: “I like order”, many candidates who are actually disorganized may answer “I agree” or even “I completely agree” to increase their level of conscientiousness and therefore their overall Nova score.

Most importantly, research (Oh, Wang, & Mount, 2011) has more recently proven that informant-reports of personality were better predictors of job performance than self-reports, which is really what matters to us at Nova.

Therefore, having found a partner like Yobs who pushes the science behind our assessment and their validity is critical to the future of our company.

A More Insightful Process: The Nova Personality Report, powered by Yobs

Thanks to the power and insights that Yobs provide, which is far deeper than what you could find from a self-reported personality assessment test, we are now able to provide our candidates with a personalized report. This personality report not only assesses their Big 5 personality traits but goes further by assessing sub traits in each of those 5 traits as well as overall strengths and weaknesses of any given personality.

Our amazing tech and design teams have also worked together to make this available in a downloadable report format which some Novas have already tested. This will be available to all candidates that go through our process very soon, regardless of whether they are accepted into Nova or not. Here you can see a snapshot of the report:

 

 

Just getting this level of insight into one’s personality should be a great reason to apply to Nova. If you are not a member yet and will like to join the community and get your report, talk to your colleagues to receive a nomination and we will love for you to apply to become a member.

Introducing our partner: Yobs

Founded in 2018 by Raphael Danilo and Federico Dubini, Yobs is a fast-growing startup that enables to record, transcribe and analyze interviews, both recorded and through live tools such as Zoom, to enable companies to make better recruitment decisions.

The Yobs API enables to provide responses to job-related questions in audio, video, or text format. Their scientifically validated assessment service then uses trained professionals and natural language processing technology to assess an individual’s communication style, work style, soft skills, and the Big 5 personality traits. This is how it works:

  • The speaker’s naturally spoken answers are transcribed into text automatically in 60+ languages. The content is then analyzed linguistically for word choice, sentence construction preferences, and more using natural language processing. Yobs does NOT do facial analysis for bias and privacy reasons.
  • Yobs uses best practices in industrial-organizational science to link the assessments to job relevance, providing pre-built target profiles and a set of questions to generate a desirable variance in responses and results.
  • Yobs’ models and datasets are constantly vetted for the adverse impacts that can arise from demographic, physical, linguistic, and accent differences between individuals.
  • Yobs’ API generates a confidence score for every report to help ensure the highest quality. Their team of expert reviewers, assisted by quality assurance tools, ensure that every report sent to our Nova members is accurate.

Happy to announce this partnership. To the stars and beyond! 🚀🌟

Leave a Reply

Your email address will not be published. Required fields are marked *

Bitcoin
Posted on November 30, 2021

ABOUT BANNING AND TAXING PROOF-OF-WORK BLOCKCHAINS: AN IN-DEPTH ANALYSIS

  ABOUT BANNING AND TAXING PROOF-OF-WORK BLOCKCHAINS: AN IN-DEPTH ANALYSIS YET ANOTHER ARTICLE DISCUSSING WHY SHOULD WE BAN (AND TAX) BITCOIN   This article is by Alessio Russo. Alessio is a Ph.D. student in Reinforcement Learning at the school of Electrical Engineering and Computer Science at KTH Royal Institute of Continue reading…

Nova Talent
Nova Talent
Posted under Nova Community
Yobs Report Nova
Posted on October 14, 2021

Using AI for Faster and More Accurate Assessments – Nova partners with Yobs

As I introduced in my previous post on Radar, at Nova we pride ourselves on having developed a technology (which we call Radar, following our galactic metaphor), that enables us to assess candidates at scale. The Old Selection Process Up until last week, the process of Radar that nominated candidates Continue reading…

How the radar process work
Posted on September 28, 2021

Introducing Radar: the Nova way to assess Talent at scale

I often get asked the following 2 questions around Nova’s assessment of talent: How do you assess talent at Nova? How do you define talent? With this brief post, I am starting a series on our assessments to bring light and transparency to them. Thus, I will showcase the level Continue reading…

E2E testing using selenium and mailtrap
Posted on September 7, 2021

E2E testing using Selenium and Mailtrap | Nova hacks to send organisational emails part 2

And how we avoided messing up the communications with our users | #NovaTech Have you ever received an email that looked something like this? Or even worse, have you ever sent such an email from your platform?  Let’s be honest, we’ve all been there (probably not as bad as the Continue reading…

Williams Aguilera
Williams Aguilera
Posted under Tech-Pills