Yes, people confront me. They argue that 5-Whys is a “good enough” root cause analysis tool. They say it’s simple to ask why five times and that people in the field can “understand it.” Well the time has come to put that argument to bed. Five Whys is inadequate. It has fundamental flaws that make it inadequate. Flaws that don’t exist in TapRooT®.
Problem 1: Confirmation Bias. Confirmation bias is a problem that all investigators have. It has been proven by extensive scientific research. People tend to jump to conclusions when solving problems. The experienced and inexperienced alike fall victim when performing unguided deductive reasoning (5-Whys). TapRooT® avoids this bias by first requiring collection of information to thoroughly understand what happened before starting to find out why it happened – thus avoiding the bias. Also, the guided root cause analysis helps the investigator consider all possibilities.
Problem 2: Can’t Go Beyond Current Knowledge. If confirmation bias isn’t bad enough, deductive reasoning (5-Whys) can’t go beyond the person’s current knowledge. One must know the “cause and effect” chain before one can find it. TapRooT® has a built-in expert system to help investigators find problems that they previously would have overlooked. After every TapRooT® class, students report success finding causes they previously would have overlooked. Case closed? Not yet.
Problem 3: Single Cause Issue. Even those who use 5-Whys admit this problem. Often people who apply 5-Whys follow just one causal chain. Since accidents are seldom caused by a single Causal Factor, they only analyze a fraction of the issues that need to be solved. They miss the chance to solve major problems. This happens all the time. TapRooT® helps investigators look for all the Causal Factors. No missed opportunities. All problems analyzed.
Three strikes … You’re OUT! That’s not all the 5-Why problems but it’s enough for me to say: “Good Bye 5-Whys!”
Category: Investigations, Root Cause Analysis Tips, TapRooT
25 Comments »
RSS feed for comments on this post. TrackBack URL







Here are the other major articles on this blog on 5-Whys …
http://www.taproot.com/wordpress/archives/1001
http://www.taproot.com/wordpress/archives/710
http://www.taproot.com/wordpress/archives/134
Comment by Mark Paradies — February 3, 2011 @ 3:13 am
Gentlemen,
After a side by side comparison investigation using our “tried and true” methods, and a TapRooT methodology investigation, Mark Paradies comments in this article are accurate.
We had the expertise of a TapRoot practitioner and lecturer to guide us through the process, and the conclusions and corrective actions that derived from the TaprooT method were superior to the results of the comparison investigation.
Effective corrective actions can now be implemented for this incident, as all the causal factors have been identified.
I would recommend this system for all those whose systems are not giving them the full picture.
Comment by Chris Rugg — March 7, 2011 @ 10:32 pm
Bias is not a condition of the method but is a personal influence that exists in anything we do. interesting to see that taproot appears to be a formailsed Ishikawa process – just another method.
please note 5 Why requires genchi Gebutsu for it to work correctly and is simiar to an incident investigation. Apollo and I-CAM are good methods also and are all limtied to the facilitator, rigour and the facts available.
intersting to see the Bias in the article
Comment by stuart — March 8, 2011 @ 12:38 am
Agree with Stuart on confirmation bias – yes, it is an issue with 5-whys, but also an issue with any other investigation methodology – the investigator brings their experience and opinions with them whatever the methodology – this is not necessarily a bad thing, but experienced investigators do need to be aware of their biases
Comment by Peter Smith — March 8, 2011 @ 5:04 am
Your comparison is between badly done 5 whys and well done tap root. If one were to flip the condition, the observations would still apply and tap root would be “struck out”. I am not a fan of 5 whys, and do not teach it when I teach root cause. It can work well to prompt questions and organize information. It does not replace the need to collect information, interview people, organize the cause and effect timeline, assess barriers, etc.
Any way, back to the first point. If one assumes a poor analyst, that person can blast through the savanah river matrix (I’m not a current licensee of taproot so I’ll not speak to its present version) and answer that questions according to their bias. I’ve seen it done under the “circle your favorite cause” approach. Contrast that with a second analyst for whom each level of five why questions forces new investigations. Strike one for taproot.
Your second point is basically still valid. A pre-existing hierarchy can force a broader viewpoint, but your first point of bias cancels it because the poor analyst ignors the questions he or she doesn’t believe in. Strike two.
Single cause: again, this is poor analyst. I pulled out my old Savanah River flow chart, took an event, and chose my cause. The tree gave me more things to consider, but in the end, (getting into the poor analyst role) I chose one. Strike three. The good analyst, using the 5 whys did not strike out, although she / he had to work harder because he didn’t have a cause hierarchy as an aid.
I think the bottom line is the hierarchy, with its predefined questions helps the analyst by giving broader considerations. As does MORT, Apollo, and the old classic, direct derivation. It does not eliminate bias, a strict use of fact and analysis does that, if it is possible to do it. it does not prevent single causes. Organizational culture tends to drive that.
Comment by Jim Gallman — March 8, 2011 @ 10:21 am
Why?
Comment by Octavio Espinoza — March 8, 2011 @ 10:23 am
Whether bias exists or does not exist in Taproot could easily be settled by experiment (or maybe it has). Give two very different groups with confirmed biases the same real or designed incident for investigation with TapRoot. Then compare the results.
I agree that 5-Why’s has no mechanism to protect against bias, but is it really the investigator’s, or is it the witness/victim/employee’s bias who is being interviewed?
Comment by Charles Lankford — March 8, 2011 @ 10:46 am
Tools don’t ensure the proper rigour – and there is nothing guaranteed out there. Gallman’s response is probably closer to the truth. If as investigators we become enslaved to a specific invesigative process, we inherit all its limitations.
A good investigation will have a team of investigators from a broad range of disciplines. They will understand the benefits and drawbacks of engineering techniques and products (or the lack thereof) such as job hazard analysis, HAZOP, MORT, PRA, etc. They will know where the relevant information is – and many times it can be in some strange places and in strange forms. An independent observer (with organizational and group behavior skills) of the investigation team is also a good thing to have where the stakes are high.
More importantly, once the team’s knowledge of the facts becomes mature, they should be able to think outside the box.
And – last but not least – seeking perfection will frustrate, but doing an excellent job is within reach.
Comment by Ron Montague — March 8, 2011 @ 10:48 am
Oh, I almost forgot! The 5-Why’s almost always are programmed to have the employee admit ‘mea culpa’ and exonerate the management system or organizational climate that contributed to the incident.
This is exacerbated due to the immediate Supervisor almost always being part of the “investigation” team, and you will have a foregone conclusion. More fodder for the “most accidents are caused by employee errors” crowd!
Comment by Charles Lankford — March 8, 2011 @ 10:53 am
I believe another short coming of 5-Why’s is it usually gets you down to what would be called the “causal factor”, and not to a true root cause. 5-Why’s and Fishbone Analysis is limited by the talent you have in the room, and knowledge. True about “confirmation bias”
5 Why’s can be used as part of developing the time line about what happened. The same is true for Change Analysis. It can find out what happened. Why it happened is usually more in depth.
Keppner-Tregoe’s old example of the leaking filter pots from the chapter on change analysis identified the use of two different gaskets for the hand ways on the unit, and it service with a particular maintenance crew as the root cause of the leak.
The root cause would be deeper in the tree. Training NI, maybe No Procedure, No SPAC etc.
Also, the tree will guide an investigator through the “why people make mistakes”. In the end, the reason we do the investigations is to prevent them in the future. Most investigations are reactive, and there is value to under stand what happened for accountability reason. However, the real goal is to prevent them from happening in the future and prevent similar incidents by proactively correcting “general root causes”.
Comment by Jason G Laws — March 8, 2011 @ 11:49 am
For those above who are trying to say that any method used badly will fail ….
That IS NOT my point.
My point is that 5-Whys (and other cause and effect based systems) have inherent flaws built into the process. These flaws exist when the tool is used PROPERLY.
That is why so many of the 5-Why examples by “expert” users (including Tailchi Ohno) are bad examples of root cause analysis.
See this bad example from Tailchi Ohno (often cited as a good example of 5-Whys by people who teach it):
http://www.taproot.com/wordpress/archives/1001
As for Jim Gallman’s comments, I suggest that he should attend a TapRooT® Course. I bet he will leave with a different perspective. TapRooT® isn’t just a list or hierarch. By design, there’s more to it than meets the eye.
Can TapRooT® be miss-used? Of course.
Is it less likely to be miss-used. Yes.
If it is used correctly, does it produce better results than 5-Whys used correctly. YES!
That’s my point. TapRooT® has methods to help it not be miss-used and processes that help an investigator produce superior results.
5-Whys is prone to miss-use by even the best investigators because of inherent flaws.
Hope that helps straighten things out.
Mark
Comment by Mark Paradies — March 8, 2011 @ 1:26 pm
I agree with Mark. The test of methods is success in reducing accidents while meeting objectives. Maybe there is some data on this?
Comment by Bill Reuland — March 8, 2011 @ 3:40 pm
Our user’s performance improvement records show the “test” results and demonstrate TapRooT®’s success.
For a sample, see the Success Stories (right column) at:
http://www.taproot.com/about.php
One more note. Even though we talk about proper and improper use of a system, the real test is the amount learned before the person was trained compared to the amount learned after they are trained. That’s where the fundamental drawbacks of 5-Whys hold people back. They are not performing tremendously better investigations after they are trained.
Best Regards,
Mark
Comment by Mark Paradies — March 8, 2011 @ 5:49 pm
There is merit in all analysis of incidents. The trick is to get people to do it in the first place. It is very difficult to use a structured system if the majority of the team is investigating an incident in their second/third language. The 5-whys can be used in any language, as long as you know the translation for 5 and whys – stick in a good facilitator and you’ve got something to work with.
In a structured predetermined selection system, if you have native speakers and non-native speakers in the team, in my experience the native speakers will rule. The causes or routes in the structured system may not be fully understood by the non-native speakers. Yet those folks, generally, are at the sharp end, where input is ‘critical’.
Managers in foreign Multinational organizations are usually, but not always, native speakers (company’s business language). They genuinely believe that they are knowledgeable in all aspects of their organizations activities – unfortunately they are seldom ‘masters’ of those activities. So really good ‘stuff’ gets missed, as those managers would not leave the ‘investigation’ to local staff alone. Expat ‘bias’.
So if you put that equation together it looks something like this:
Preprinted/screened selection process + Expat ‘bias’ + non-native speakers + native speaker rules = potential for ‘lopsided’ analysis
Now if all participants are ‘selection process’ first language speakers the argument can and most probably will change – but for the past 32 years I have never worked in a company where all employees had the company’s business language as their ‘first language’.
Next year I will have been working for 50 years, the first 17 in the British Royal Navy where incidents ‘didn’t happen’, so I can’t count those in the equation, even though we were all native English speakers.
Comment by Roger H. Willmott — March 9, 2011 @ 4:40 am
No question TapRoot is a far superior RCA tool. So What. 5 Why is just a subset or child of Why Tree which does look at and attempt to find all root causes. Back to 5 Why. You have to be willing to accept the fact that as you work down from the top of the Why Tree you pick the maybe or what appears the most significant cause and only work down that one branch. It is not meant to find all root causes it is meant to find 1 that will prevent or reduce the event from happening again. If I have a serious complex event occur I would always prefer TapRoot. Why Tree works well enough on 70 or 80 percent of the evens I encounter as long as the team and facilitator have the experienced necessay to verify and follow logic. 5 Why is a way to address an event but not wear everyone out addressing every possible cause. Also 5 Why does not ask 5 Whys. It might take 3 or it might take 10. Do the math over 100 investigations and the average will be 5.
Comment by Roger Harris — March 9, 2011 @ 8:18 am
Hi Roger W,
We have lots of international clients and have translated TapRooT® into Spanish, Portuguese, German, and French.
But the main techniques used with those who are “untrained” is the SnapCharT® and it is always used in the native language. Thus, I don’t see where the investigation will go astray because of language issues if people use TapRooT®.
Best Regards,
Mark
Comment by Mark Paradies — March 9, 2011 @ 10:47 am
Hi Roger H,
I have been debating the “time savings” vs “quality” issue in my mind for many years. My most recent thoughts were published in this article:
http://www.taproot.com/wordpress/archives/18906
To summarize, if it is worth investigating, it is worth investigating well. If it isn’t worth investigating, don’t waste your time investigating it.
Of course, this assumes that one can tell what is worth investigating and what isn’t.
Best Regards,
Mark
Comment by Mark Paradies — March 9, 2011 @ 10:51 am
We often talk about using TapRooT® proactively to avoid the disaster rather than investigate it. I am a firm believer in the TapRoot and HPES processes. On several occasions I have done TapRooT® on minor problems; minor injuries or administrative errors that had very little potential for serious consequences, but throught the TapRooT® thorough investigation process we have kicked up “significant issues” that were time bombs waiting for us. In an investigation of a minor administrative error in a Lock Out Tag Out job, we uncover a completely separate problem with a Tag Out Lock Out listr that was an industrial explosion waiting to happen. The TapRooT® process has paid for itself many times over, but I also agree with some of your other readers. It’s only as good as the people using it. You can do wonderful things or Great damage with any tool. That’s why the TapRooT® TRAINING is what’s vitally Impiortant. More so even than the tools themselves.
Comment by Lee Dobry — March 9, 2011 @ 11:54 am
You say “if it is worth investigating it is worth investigating well”. I say if you can get 50 supervisors to look at 100′s of events and complete 100′s of actions every year you have a workforce that is enlightened not tired of being in another investigation meeting. They feel empowered to make a difference quickly so do their best to do so. You only need to control the actions and give feedback on their quality and effectiveness. You also build a tremendous data base that can be used for trending and analysis.
Comment by Roger Harris — March 14, 2011 @ 9:38 am
Roger
Yes – Supervisors need to do investigations and need to do them well.
Supervisors do not have majic answers for difficult problems. They need a robust root cause analysis tool as much or more that engineers or managers.
Also, hundreds of poor investigations don’t produce a tremendous data base, they produce garbage in = garbage out.
That’s what I frequently see at places using 5-Whys.
Mark
Comment by Mark Paradies — March 14, 2011 @ 11:31 am
Hi,
After completing the 5 day taproot in Scotland last week, I can honestly say ”WOW”. Im really impressed with the technique and methods to investigate incidents and gathering/clarifying fact before breaking causal factors down to root cause.
5 why is limited and of course bias in terms of scope – previously i have integrated cause and effect then 5 why to break down to root cause but agian this is purely subjective.
Thought id share this feedback – On this subject, I am in the process of executing my first taproot as team leader starting tomorrow – no time like the present!
Cheers,
Billy
Comment by William R — March 15, 2011 @ 6:53 pm
Thanks for the course feedback.
After a couple of years let me know how many accidents have been avoided and lives have been saved. That is the best measure of success.
Mark
Comment by Mark — March 15, 2011 @ 7:17 pm
I’m a bit late on this string but feel compelled to add my bit to a very interesting discussion.
I have to agree with Roger Harris on both of his comments. Further I don’t think you can compare the basic 5 Why with a much more complex Taproot. At my facility we use both 5 Why and Taproot as investigative methods based on the severity of the incident (a subject on its own).
Is it fair to categorically slate one method over another? I don’t think so. As many of the comments indicate it depends on the people who is at the helm of the investigation process that will determine how effective the outcome of the investigation will be.
A fundamental part of preventing the incident from recurring or to significantly reduce the potential for recurrence, is the effectiveness of the resultant solutions and crucially the implementation thereof. If the solution meets the SMART criteria I don’t think the methodolgy matters. How often has an investigation been completed with recommendations taking forever to be implemented or never verified and vaildated?
Comment by Wayne Williams — May 11, 2011 @ 9:40 am
Good day gents,
There are very interesting comments on this article. I agree with the author about the facilitor’s skills when we need to run an investigation, we must be objective. However, I cannot see how he/she is not under bias on Tap-root. In the company where I work we use the ICAM Method to investigate serious incidents, where one step in the mehodology is to use a Timeline and 5-Why’s. If you have an unexperienced investigator, obviously you will not have accurate results, but it’s different with a skilled one.
I agree, it’s not an easy tool to look for findings on an investigation, but I would not like to use a front-line worker running a serious and complex investigation, but I’m not saying his/her participation would not be useful. The “facilitation” of an investigation must be done by people who knows what to do or to look, especially in big operations like mine sites.
Finally, the author’s arguments against other methods and/or tools for incident investigations should not involve a kind of a marketing campaign for Tap-root, that’s not serious. Also, I don’t remember if James Reason (the gurú of the swiss cheese model) is against 5-Why’s, I don’t think so.
Cheers.
Comment by Diego Barahona — July 12, 2011 @ 9:06 am
James Reason built a model to find root causes based on Don Norman’s work. See what Reason said here:
https://www.distancelearningcentre.com/access/materials/cog_psych/James_Reason_Lecture_on_Action_slips.pdf
Then look at a complete 5-Why critique at:
http://asq.org/quality-progress/2010/04/quality-tools/under-scrutiny.html
Comment by Mark Paradies — July 12, 2011 @ 11:16 am