This article was written by Professor Marko Kesti and it was originally published in Finnish here. The following blog post is a translation of the origial article from Finnish to English.
There is a serious mistake in employee surveys that hinders the development of employee productivity. To illustrate the problem, let’s look at an employee survey with three questions:
Question 1: I find my work very meaningful
Question 2: Our work processes are efficient
Question 3: Fair treatment
The survey was conducted in two workplaces A and B. Case A is a workplace where work is not perceived to have a significant meaning and work processes are not very effective. Nevertheless, the workplace is good in the sense that employees are treated fairly. The grades for case A are: Question 1: 60%, 2: 60%, 3: 100%. Thus, an average is 73%.
According to this result, there is not that much need for improvement. Yet, the result of 73% is incorrect because there is significantly more development potential. This is due to the fact that the employees do not pay any attention to the fair treatment because it is already at a satisfactory level. Therefore, it should be excluded from the results analysis as being irrelevant. From a scientific point of view, ignoring unnecessary variables and focusing on important issues is called applying Bayesian theory.
The problem is particularly visible in extensive employee surveys in which the things that are satisfactory and therefore irrelevant, augment the result. According to the Bayesian theory, the workplace A has an average of (60%; 60%; 100%) = 60%. Thus, there is a big difference compared to the traditional average score of 73%, which conceals the development potential. It should be stated that according to the QWL index, the result is 55%.
Let's look at another case. In the workplace B, there occurs bullying from the supervisor. Employees feel that their work is meaningful, and when the supervisor is absent, their work is efficient.
The results of the employee survey are the following:
- question 1: 80% (meaningfulness)
- question 2: 80% (processes)
- question 3: 20% (fairness)
The fair treatment receives a very bad rating (20%) and now, according to the Bayesian theorem, it becomes a conditionality. Based on the average calculation, the result is an average of (80%; 80%; 20%) = 60%.
However, the result of average calculation is misleading because the good grades override the bad grades. In real life, people focus on the unfair treatment - a threat that is probable in their workplace. Question number 3 (fair treatment) becomes critical, so that other questions do not really matter (when the supervisor is present). According to the Bayesian theory, the result is 20%, since questions 1 and 2 are ignored, so the average of (80%, 80%, 20%) = 20%. Thus, there is a big difference compared to the traditional average score of 60%, which conceals the development potential. It should be also noted that the QWL index yields 25%.
Traditional employee surveys are coming to an end. Because of various problems, they do more harm than good for the organizational management. The biggest disadvantage comes from the fact that they conceal the potentiality of increasing employee productivity. It seems that humans makes decisions according to the Bayesian theory.
A human analyzes probabilities and relevancies. If something is insignificant, you do not need to use your energy to think about it. Our QWL index, developed at the University of Lapland, applies the Bayesian theory and considers also the nonlinear effect of various questions on performance. The example above demonstrates how the QWL index can give a more accurate result in both cases.
From the point of view of utilizing artificial intelligence, the new QWL theory is very useful, even groundbreaking in many ways. Therefore, I have thought of writing more about this inspirational topic. The development of QWL can be explored further in training, in which artificial intelligence assisted simulation is being used.