Our purpose is to make the world more equal through fair recruitment practice. Testhub is a recruitment tool for automated screening of job applicants that predicts job performance and minimizes cognitive bias in recruitment decisions.

Cognitive bias has been and continues to be a huge setback in creating a truly equal, diverse and inclusive workplace. Organizations’ attempts to achieve diversity are obstructed by unconscious bias from the point of recruitment. Studies have shown that such biases can hinder equal opportunities, especially for women in terms of selection and progression to high-level management and leadership roles. Research has also shown that a diverse workplace is more effective and that a diverse organization performs better. Our tool has been designed to be a solution to prevent such biases in screening of candidates.

Testhub is a tool based on scientific consensus. One of the most comprehensive and most cited meta-analysis in the scientific field of personnel psychology is a key ingredient in our product. Meta-analysis is the statistical procedure for combining data from multiple studies. When the treatment effect (or effect size) is consistent from one study to the next, meta-analysis can be used to identify this common effect.

This meta-analysis is a research paper that sums up the accumulated research findings on what traits and methods are most valid in selection of employees over a 85 year period and it has some clear implications for how one should go about recruitment. (Schmidt & Hunter, 1998)

It is clear for the research that many if not most tests and methods have low incremental validity which means you are often better off by not using them. We only focus on the traits and methods with proven predictability to make sure you only test for things that actually matter for work performance.

Some of the authors republished this research article as a working paper including the last 20 years research in 2016. Most of the conclusions are the same, but one aspect has increased in importance as general mental ability and degree of integrity combined has the highest predictive validity for work performance. And that general mental ability measures combined with a structured interview is the most pragmatic method. (Schmidt, Oh & Shaffer, 2016).

A combination of the best method and the strongest indicators that someone will perform in any job. Our general performance indicator (GPI) is calculated based on a short questionnaire the employer and the candidate fill out. The candidates' results will be shared with the employer before identity in order to minimize bias in deciding who gets an interview.

The employer will have an anonymous list of candidates ranked after likelihood of work performance and only when they choose to book an interview will identity be visible.

If the employer agrees, we will also share the employer's test results with the top candidates because this will give candidates insight into the likelihood of the performance of their potential new leader.

We also match our employers’ attitude towards autonomy with the candidates’ need for this. This is because autonomy is a predictor of turnover intention and an important factor in intrinsic motivation and thus employee retention.

A big problem for absolutely any human decision-making is the numerous subconscious biases that guide our behavior. This entails that we as human beings have an innate tendency to make mental shortcuts when we make decisions.

Decades of psychological research have shown the limitations of intuitive expert decision making. Most often this error in making choices happens without our knowledge. It's part of how our intuition works and one of the most prominent scientists within psychology, Daniel Kahneman, strongly believes most experts make great intuitive errors.

Being an expert actually makes you more exposed to overconfidence bias than a normal person. Overconfidence bias makes you more likely to ignore data and follow your gut feelings. This is just one example of the complex process of evaluating others and our and experts' blindness to these mental traps.

Daniel Kahneman is greatly inspired by psychologist Paul Meehl, who has published one of the most controversial books in psychology in his Statistical prediction VS Clinical intuition (1954). This illustrates 20 studies with a strong correlation in favor of mechanical prediction over human expert intuition. This brings us to a key point about our service: The formula we use to measure the candidates and employers is mechanical prediction where the variables are traits with the best predictive validity for performance combined. This aspect alone will drastically increase the likelihood of hiring a better candidate.

We have made a tool that is designed to minimize the known biases in the recruitment process. By using this method you will decrease the bias in your process and increase validity. This tool will get you better and more diverse employees partly because you will not know their identity until you decide who gets an interview from an anonymous list. There is just to much room for unconscious discrimination in knowing someone's identity beforehand so we help you judge on the things that matters according to the science.

The GPI- General Performance Indicator

This is a formula that predicts performance in any job by combining variables that all have high predictive validity for work performance. Formulas like this have been proven to significantly beat an expert's intuition in 60 % of over 200 studies done since the 1950s. In the other 40 % experts and formulas are almost equal making formulas much more reliable than experts. (Kahneman 2011).

Another important theorist our product is inspired by is Robyn Dawes (1979). He made a significant contribution to statistical prediction by uncovering that a simple set of scores with some validity to the purpose of predicting with equally weighted values will be just as accurate as the multiple regression formula.

Use of hiring methods with increased predictive validity leads to substantial increases in employee performance as measured in percentage increases in output, increased monetary value of output, and increased learning of job-related skills (Hunter, Schmidt, & Judiesch, 1990). Testing for relevant traits is better than just interviewing and in combination you get the best result. (Schmidt, Oh & Shaffer 2016)

The Autonomy Matcher

This is a matching test that matches the candidate’s need for autonomy with the company culture for independence. Autonomy was found to be the strongest factor in retaining employees in EY and HR Norway's 2018 report: 80 % of the organizations who report they give high degree of autonomy also report succeeding in keeping key employees.

Clients will be provided with guidelines on how to perform structured interviews with questions to score. By implementing the structured interview and numeric score method employers will significantly increase the likelihood of getting the best employee by increasing the validity of the process even more. For the optimal result we hope employers perform and score the three structured interviews that they are provided, and that there are three different people in the company that performs these interviews, and that these people should not know any interview results beforehand. Having then three people’s almost objective observations and score of the candidate we can calculate the average score. By doing this we will minimize exposure to a relevant statistical phenomenon called regression towards the mean. This entails that the average score of three interviews will be the most true representation of the candidate since people tend to fluctuate under and over their average performance thus will giving too much weight to one observation decrease validity. (Kahneman 2011) Again this method will increase the likelihood of a solid estimation of the potential employee. After this, all employers have to do is to choose one of the top candidates and they can be confident that anyone chosen with above average scores will be able to perform in the job.

Heuristics are cognitive rules of thumb that guide us in our decision-making. These mental shortcuts steer how we perceive and make sense of our reality. According to Kahneman, Slovic & Tverssky (1982) will the reliance on heuristics cause predictable biases in predictions. In the interview situation, a candidate is analyzed by the recruiter who uses the representative heuristic. The heuristic comparison is based on the interviewers previous experiences rather than logical analysis of candidates abilities.

People rely on heuristic principles to reduce complex tasks of assessing probabilities and predicting values to simpler judgmental operations. In general these heuristics are quite useful, but sometimes lead to severe and systematic errors such as biases. (Kahneman & Tversky 1974)

A cognitive bias is a systematic error in thinking. That means you are not aware that you favour or disfavour someone new because they look, behave or smell like a person from your past. You may be relying on a cultural stereotype or you could be influenced by several of more than a 100 documented cognitive biases. This is quite a mind bender as the dominating perspective that still guides our world is that people are rational.

We are in others words quite inconsistent in making choices. In fact we are predictably unpredictable as we know bias happens to all of us and influence how we think and make choices. It is also hard for anyone to predict long term results in complex low validity environments. Which is most situations, especially choosing another human to work for you. Simple formulas do better than gut feelings in these situations and should therefore be implemented in making choices. (Meehl 1954)

People are not good intuitive statisticians and our brain is rigged for jumping to fast easy decisions, and not to bother with gathering and analysing the data at hand. Kahneman's research shows that professors with extensive experience with statistical methods fail to apply those methods in evaluation of others on the spot. It's just how everyone's brain seems to be wired after a few slow million years of evolution. According to Huczynski (2004) will the biases of the human decision process operate at the subconscious level and have a powerful impact on an individuals choices and actions. According to Kahneman (1974) will people and experts not be able to adjust for subconscious bias even if they know what they are. They happen automatically and without our conscious access. When we like or dislike someone without knowing them intimately we make up a coherent, casual story of why this is.

We often evaluate others and our feelings towards them based on their or our subjective confidence. People that are high in confidence are judged as highly skilled. This is not true. Subjective confidence is just a feeling. The strength of the feeling is based on how coherent the story that we tell ourselves is in a given situation. If the story seems coherent confidence is high. This is disastrous for recruitment as you can make a coherent story out of very little information and sometimes very unreliable information. People can be very confident with very little reason. Confidence is therefore a bad diagnostic of when you can trust someone or when you can trust yourself. (Kahneman, Slovic & Tversky 1982)

Cognitive biases are common to all people in all cultures and are hard coded into humans after millions of years of cognitive evolution. Many of these gut feelings and instincts were great to have when we all lived in the forests and meeting someone from a different tribe could mean life or death. Now as the world has become more complex than our cognitive development stage has capacity to deal with, many of these are mental traps that make our decisions worse.

Please contact our CEO if you have any questions: Jens@testhub.tech