Hiring can be a crucial moment for the life of an organization. Managers find themselves in front of an amount of candidates motivated to enter their company and they have to choose who among them will get the boarding card. This choice is not irrelevant. Quite the opposite: the ability of people to complete the task they are paid for – their performance – varies consistently and the economic consequences for the companies are proven to be extremely relevant. Some researchers (Schmidt & Hunter, 1993) were able to quantify the gains of a well-managed selection: For the company the economic difference between a really good employee and a low performance one is on average as large as the 80% of the salary paid for that job! If you try to multiply this 80% for the total cost of the personnel in your company you will easily realize that selecting the best available people for a given position is critical. The problem is: how can we do it? For decades psychologists believed that each company and each position would need its peculiar instrument to assess the fit of the candidates. But starting from the 70s new statistical tools allowed HR scholars to discover that actually some tools used for the personnel selection ork better than others to predict the performance of future employees in almost every organization. Remember R In scientific literature the letter “r” is the conventional symbol to represent validity. What is interesting for an HR professional is that exists a particular type of validity that measures in an objective way how much an instrument is good to select a performing employee. It is called predictive validity and it measures how much the scores obtained in a test can predict the future performance of an employee. The score of “r” range between 0 and 1. 0 – or a score close to 0 – means that the method doesn’t give any information about the ability of the candidate and that you are just trying to see the future in a crystal ball. With score of 1 you could predict with absolute certainty how much an employee will perform only using that method. In the real life a score of 1 is impossible to find. Methods that have a higher score than 0.5 are considered really reliable. Management is a statistical science and the intuition of a sensitive professional will always have its role. The best methods Schmidt & Hunter (1998) analyzed tons of data collected in 85 years of psychological research to produce an exhaustive and reliable research about the predictive power of selection methods. Three tools have an R value higher than 0.5: work sample (0.54), General Mental Ability Test (0.51) and structured interview(0.51). Be aware: interview is a great method only if it is well structured and systematically scored. A simple unstructured interview, although its costs are smaller, has a predictive validity of “only” 0.38: Good, but not excellent. peer rating (0.49) and job knowledge test (0.48) also score really well. Integrity test (0.41) and conscientiousness measures (0.31) have smaller predictive value, but they work really well when combined with cognitive measures. An Assessment center reasonably predictive (0.37) and it is more effective than work samples for managerial positions in which personal core competencies are more important than the ability of performing a specific task. Biodata (0.35) are also good measures. Last and least The research of Schmidt and Hunter also pointed out that some methods, although used in the HR practice, are not useful for selecting personnel. So it’s better to avoid selection processes based on interests (r= 0.10) or years of education (0.10). Job experience also rates surprisingly low (0.18). In fact an experienced worker will have a better performance only in the first months or years of his work for the organization. It was tested that after 5 years occupying a position an employee reach is best level and further experience doesn’t have any effect on the performance. Anyway selection based on job experience can be useful when you need an immediately operative employee. It’s not different than a random selection any process that involves graphology (0.02) or one that is based on the candidate’s age (0.01). So, next time that a certified graphologist try to convince you that his skills could help you to hire the best employees, you are entitled to push the expulsion button or to free the dogs. Takeaways for your practice Some further words should be spent on the topic to assist you in the choice of the selection method. First of all, not all the considered tools are helpful in every situation. Work samples and job knowledge, for example, can’t be used to select candidates at the entry level that still need to learn their jobs. In these situations intelligence tests, integrity tests, peer ratings and interviews represent the most reliable options. Furthermore, in choosing your selection methods you should consider the costs. Usually purchasable paper and pencil tests represent a cheaper option. If the expected performance of different employees is not really different (for example, when the job is really easy) or if the number of candidates to possess the minimal requisites is almost equal to the number of people to hire it’s not convenient to spend too much money on the selection procedures. Finally, although job performance is usually considered as the most important characteristic that a selection method should predict, it is not the only one to matter. Integrity, organizational citizenship and turnover, for example, also have a strong impact on the organizational life and they can be predicted by different methods. What about you? What selection methods do you use? What new technique would you like to try if you could? Remember to follow us on LinkedIn, Twitter and to subscribe to our newsletter to receive all the articles we are going to publish on Recruiting and Selection, and share this information with your network by clicking the buttons below! References Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological bulletin, 124(2), 262. You can find the original article here.