SRC Forum - Message Replies

Forum: Reliability & Maintainability Questions and Answers

Topic: Reliability & Maintainability Questions and Answers

Topic Posted by: Reliability & Maintainability Forum ( )
Organization: System Reliability Center
Date Posted: Mon Aug 31 12:47:36 US/Eastern 1998

Back to message list Show all replies Topics List About this forum
Original Message:

Posted by: C
Date posted: Mon May 28 11:01:41 US/Eastern 2001
Subject: Confidence Levels
Could somebody explain confidence levels for me. What does it mean when it states a 90% or 60% confidence level and how do determine the confidence level to use? How do the results change for these confidence levels? Thanks for any help given.

Previous reply

Anthony Coppola Subject: Confidence Limits
Reply Posted by: ( )
Organization: RAC
Date Posted: Thu Jun 7 8:42:05 US/Eastern 2001
When you measure reliability (failure rate or MTBF), you can draw a distribution around your measurement representing the probability that your measurement came from a population with a true reliability of any given number. A confidence limit marks an area of that distribution. For example, a confidence of 80% in an mtbf of 20 to 40 hours means that 80% of the distribution will fall between 20 and 40 hours. Thus there is a 20% chance that the true value of MTBF is outside the range 20-40 hours. We say we are 80% confident that our MTBF is in the range or that we take a 20% risk of error when we say that. In testing, confidence is one minus the risk of error, based on a MTBF stated as desired or, more often, undesired. Thus a 90% confidence means that a product with a true MTBF equal to a value we defined as bad has only a 10% chance of passing the test. In confidence limits higher is better, but with drawbacks. A 90% conficence limit is wider than an 80% confidence limit in measurement, and, in testing, a 90% confidence test takes longer than an 80% confidence test. You can use the highest confidence that you can afford or the lowest confidence that makes you comfortable. I recommend RAC product "Practical Statistical Tools for the Reliability Engineer," which I wrote. It answers your question in detail.

Reply to this message