 |
|
 |
|
| Home > Statistics > Process Capability | Search: | for |
|
|
|
|
Human Process Capability - Defects and Errors
It's your job as a quality professional or business manager to create processes that are effective and efficient. In other words, to create processes that are capable of performing at specified levels and that reach their entitlement. It is an evolutionary improvement path from an ad hoc approach to a mature, disciplined, standardized process that fuels enhanced business performance. But how far can human processes evolve? We know that if a machine is stamping out widgets, we can use a control chart to track factors that will allow us to control the output -- to create a process that approaches Six Sigma. Can the capability of a human process ever approach Six Sigma? From some of the data presented below, the answer would appear to be "no". If a process produces 4 defects for every 1,000 opportunities (see below data), the process sigma would be 4.15 -- a far cry from 6 sigma. So can we then say that people are innately incapability of producing defect and error free products and services? I'll let you make the final call. The topic of human entitlement and capability recently came up in an iSixSigma discussion thread. It's a fabulous topic and the thread presents data that all of us should be aware of as we design and fix our current human processes. This is the first time that I've seen information published on the subject. Thank you to all that posted their thoughts and knowledge. If you would like to make an additional comment, just press the "Post A Reply »" button. "For visual inspection tasks, Fitt's Law applies. It basically says that the ability to perceive is proportional to the size of the object in the visual field. That, added to some human factor fatigue issues, says that the best you can catch over a typical 8 hour industrial shift is about 80% of the infrequent or small defects that actually occur. You can do better for short periods of time, or by using techniques to magnify the defects, or by using techniques to assure systematic multiple scans of an area. You also get better if you have frequent breaks to change focus, or set the task time for such watchfullness quite short, to eliminate the visual fatigue effect. So 95+% of defects will be detected visually under the best possible circumstances. For small motor tasks, it gets better, unless you have extremely stringent criteria for the task. That's because there's usually both visual and tactile feedback, and humans are self-correcting on the fly. So a typist can get a few minute bursts of 150 words per minute typing error free; roughly 1 error or less in a thousand keystrokes. That's for transcription typing in an excellent environment. More typical of people typing, say, stock numbers or measurements on the plant floor is one error per 100 characters. For large motor tasks, it varies all over the map. In baseball, both pitchers and hitters are considered very good if they get more than half of their large motor tasks defect free. But for throwing 1 kg. grenades, with training, typical recruits can get them to land in a 4 foot diameter target circle 100+ feet away better than 90% of the time. Of course, the grenades often slide off-target after landing and before detonation, if the timing wasn't perfect. But given the risk to the thrower of being late, this was deemed to be acceptable performance by the people writing the specifications to graduate from that training. I have a reference book, but not handy, that summarizes lots and lots of human factors studies, from which the detailed info on similar tasks (mostly large motor, some cognitive, some visual and aural perceptiveness) can probably be determined. As you can see from the grenade example, many of these studies were originally conducted by or for the US military and or for utility companies. In determining the human entitlement, you will need to pay lots of attention to your process and defect definitions. In for example, the frequently used 'better than Six Sigma' example of pilots landing planes, the specification for an acceptable (neither short nor long) landing touch point is set deliberately quite wide in terms of the braking distance needed for a plane, and the overall length of the runway. They don't call it a reportable or defective landing unless it's outside of those relatively wide, generous-safety-factor parameters. So the airplane landing human process entitlement is really great - better than 99.99%. And a minor defect - a short or long landing - still leaves a pretty large amount of room for the plane to be unharmed. So that airlines truthfully claim to have a 'much better than Six Sigma' passenger safety process. The techniques of Poka Yoke / Error Proofing can make a very large difference in how many defects the humans add or delete from a system." Posted by: Anonymous "After reading your message, I researched Fitt's Law and it's about what I suspected. The analogy that comes to mind is a person dropping marbles from a height of 3' into two circles drawn on a floor. If one circle was 4' in diameter and the other 4" in diameter, one would expect more marbles dropped outside the 4" circle (defects) than the 4' circle even though the number of opportunities are the same per circle. The rate of defect one may find is tied closely to the level of difficulty of the job being performed (in a nutshell). For this particular process the payback for continuous improvement is very small from a both a financial and customer satisfaction standpoint. In lieu of other larger payback projects focusing improvement time in this area is unwarranted. So it leaves the process champion wondering, how good should we expect to be through training, coaching, feedback, expectations, etc, if the process, from a data entry standpoint, remains as is?" Posted by: Chris "I know that DuPont as a part of its extensive safety efforts uses research on human error rates. This includes the rate of errors when a human is under stress (e.g. in an emergency situation, how likely are they to push the wrong button). DuPont presents the data on a 'demand' basis. The rates range from once per demand for the operator failing to react correctly in the first 60 seconds after the onset of an extremely high stress condition (such as a very serious process incident) to once in 10,000 demands for selecting the wrong key-operated switch under normal conditions. General errors of commission (e.g. misreading a label or document and then acting incorrectly) are reported to be 3 per 1000 demands - in the range you mentioned. I pulled these numbers out of DuPont's Process Safety Management Reference Manual. Perhaps someone in the DuPont organization could get you more information on these. They have a bibliography, but that covers all of the expected equipment failure rates as well and I can't cull out which reference applies just to these human error rates." Posted by: Mike Beauregard "I am not aware of any research on the issue of human entitlement - however, I am currently in the process of closing a project that has involved a human process - the closing concerns per thousand is 4.5/1000 (as the previous postings suggested) - I deemed at the time that this was the entitlement, so it conforms to what you have heard. The problem I have had is maintaining the level of concerns per thousand, as the controls that you can put in place are pretty limited (e.g., standardized work; error proofing; standardized tools, etc.), and these do not ensure the same level of consistency as controls on a machine / tolerance." Posted by: Simon Reproduction Without Permission Is Strictly Prohibited – Copyright Requests Publish an Article: Do you have a Six Sigma tip, learning or case study? Share it with the largest community of Six Sigma professionals, and be recognized by your peers. It's a great way to promote your expertise and/or build your resume. Read more about submitting an article. "The Bottom Line" Links
Download the iSixSigma Toolbar for 1-Click access. Search Your Way. Everyday. Without Delay.
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
| Home | Discussion Forum | Event Calendar | Job Shop | |
 | |
| Link To iSixSigma | Rate This Page | Report A Problem | Free Content For Your Site | Submit Article For Publishing | |
| |
| Terms of Service. ©2000-2008 iSixSigma. All rights reserved. v3.0lb, 0.6-C-246 |
About iSixSigma · Contact Us · Privacy Policy · Site Map. |