This past week I’ve heard from a few that have taken the ASQ Certified Reliability Engineer exam. The common comment was on the use of MTBF and the exponential distribution. It seems to emphasized.
That is bad.
MTBF, as those that read this blog regularly, know my opinion of the use of MTBF and the associated assumptions. In short, they lead to bad decisions due to an oversimplification of representations of the data.
The body of knowledge does mention many other distributions and not only exponential. Yet, it seems it is easier to write questions for the exponential or when using MTBF than others. Or the filtering of questions does not take into account the weight MTBF is being given in the exams. Or, those that craft the exams feel that MTBF is of value and should be well known and well tested by those testing for the CRE.
What ever the case – it should stop. Stop using, encouraging, testing and preparing for the use of MTBF. Reduce it’s importance and reduce it visibility. Maybe then our profession will start to learn and use on regular basis those tools and techniques that are truly useful. If one is faced with time to failure data, yes it is easier to calculate MTBF rather than a Weibull fit. Yet, we should be using Weibull as a minimum and the appropriate approach when justified.
We should not reinforce the notion that we can simply assume constant failure rates and drive on. In my opinion we need to break this cycle and do it now.
So if you are writing question – don’t use MTBF or exponential. If you’re crafting the exams – minimize the use of MTBF and exponential. And, if you’re taking the exam – provide feedback about the overemphasis of MTBF which isn’t appropriate as a reflection of the BoK.
Together we move this a bit and help the entire industry.