The truth behind why performance ratings are being scrapped
About the author
Stephane Michaud, PhD, Senior Director – Consulting, Human Link Asia, is host of HRM Magazine's The Academic View forum. He regularly contributes articles and discussion on the ways science can help HR become even more effective in its role.
Psychometrics tests have been used to measure psychological attributes for over a century now.
Inspired by Charles Darwin’s Origins of Species, Alfred Binet developed one of the first psychometric test designed to measure intelligence – the Stanford-Binet IQ test in 1905. The Army Alpha test was then created by Robert Yerkes in 1917 to systematically evaluate the intellectual and emotional functioning of soldiers and was one of the first attempts at employment testing.
Since then, countless assessments have been developed and validated, not only to measure intelligence or "general mental ability", but various psychological attributes including personality. In addition, myriads of studies have demonstrated that reliable and valid psychometric assessments accurately predict job performance (mostly supervisory ratings).
In the face of such overwhelming evidence, one will be forgiven to ask: "In this day and age, why are employers still reluctant to use psychometric assessments in the selection and promotion of employees?"
It flies in the face of rational argument. Not to mention that within the substantial group of employers using assessments, many are still using them wrongly. (i.e. administering them after selection interviews rather than at the beginning as a screening device, which is its intended use).
Some governments (i.e Canada) currently use psychometric tests for promotion, but by and large, they are shunned by all but a few employers for career development and succession purposes. The cynical view, aptly fitting recent socio-political events, would explain this state of affairs by hiring managers’ preference for "alternative facts".
However, I would like to propose that there are two major reasons as to why employers would rather select their employees using unstructured interviews rather than with properly validated means.
The first reason, although not consciously pondered by most employers, is the legitimate concern that the measure we are trying to predict, namely job performance, is flawed. You see, psychometrics assessments are useful to the extent that, not only the predictor is reliable and valid (i.e. psychometric tests measuring cognitive ability and personality traits), but also that the criterion is reliable and valid.
Now this is the kicker: we can predict fairly accurately whether a candidate will behave in a diligent, agreeable or extroverted way. However, the issue is that we are trying to use psychometric tools to measure psychological attributes in order to predict a measure with notoriously low reliability and validity (i.e. supervisor ratings). With so much “noise” in the criterion, the predictive power of our assessment tools is greatly diminished.
Indeed, it has been demonstrated that objective performance measures have moderate to low correlations with supervisory ratings. Undoubtedly, this reflects supervisory biases involving matters of the heart which have little if anything to do with performance.
However, as a number of consultancies have found out through research over the last few years, supervisors are not only rating the task performance of an employee but also their “relational” performance as a member of their team. Therefore, it’s no accident that rating intentions correlate much better with supervisory ratings than anything written in the much-maligned annual performance plan.
So until we find better measures of employee performance reflecting these psychological realities, the traditional performance management cycle will continue to frustrate everyone involved. Current performance systems reflect the age-old conflict of interest between the desire of the supervisor to make employee decisions based on their intentions, and the needs of the employee for predictability and protection against arbitrary judgements. No wonder, over a third of companies in America are getting rid of their performance system. In summary, flawed criterion measures blunt the efficacy of our predictive assessment tools.
A second related and poorly researched question involves the assumption that supervisors’ primary motivation in hiring employees is to improve or maximise job performance. Although employee job performance is in the interest of the company as a whole, I would argue that it may not be the most important factor on the minds of hiring managers.
Unfortunately for organisations, supervisors are not the rational beings that science would like them to be. They have a vested interest in ensuring that they choose employees who, they feel, will be able to get along with them, and to a lesser extent, get along with the team.
In other words, psychometric tools are designed to predict job performance, while supervisors are trying to predict whether candidates will become good workmates.
To a certain extent the trait “agreeableness” could help managers in this regard. However, I believe that a more promising area worth exploring lies in measures of organisational and team fit. These measures do exist but are not as widely validated and utilised as psychometric assessments. Until we address this fundamental interest of hiring managers, many will continue, for better or worse, to resist psychometric testing.
Stephane Michaud, one of HRM Magazine's forum hosts, regularly shares why research and science are imperative to the implementation of effective HR strategies.