woensdag 21 september 2011

Discussing Heinrich, Part 0: Introduction

Some time ago there was a thread on the EHSQ Elite mailing list that lead into a discussion of Heinrich’s work (mainly the triangle) to which I wrote a rather lengthy reaction. At that time I had neither read Heinrich’s book (of which I own the 1941 second edition and the 1980 fifth edition that has been reworked by Dan Petersen and Nestor Roos) nor “Heinrich Revisited” by Fred Manuele that was mentioned in the ESHQ thread. Well, all of this went high up on my list of priorities and after reading all three of them, it’s time for a critical discussion.

For starters a minor disclaimer: what follows is by no means a scientific discussion, but rather a more or less structured assembly of notes that I jotted down while reading and trying to sew them together into a more or less coherent evaluation of these three works. I might consider a more scientific and theoretically grounded reworking at some point in time. All depending on moods and time available…

I’m going to do this in four parts. The first three parts will take on the separate books (with a possible crossover, or two). In the fourth part I will try to sum up some things and come to some kind of a conclusion.

To the next part of this article...

Discussing Heinrich, Part 1: Heinrich '41

This is the second edition of Heinrich’s famous book, and from what I understand from the preface it has been reworked to a certain degree, yet I cannot say what the difference from the original is. Let’s go through the contents with some remarks underway:

Chapter 1 discusses Heinrich’s principles of accident prevention. One of his key themes is “hidden costs” of accidents that pop up already on page 2. The fundamental principles for accident prevention, according to Heinrich, are first listed 4 pages later: 1) creation and maintenance of active interest in safety (we could call this commitment, I guess), 2) fact finding, and 3) corrective action based on those facts. Quite common sense and we’re probably all in agreement. He goes on to discuss the fact that work on safety should be both top-down and bottom-up, but safety is mainly a management responsibility. Also Heinrich states that the work should be done methodically, based on scientific principles as he says, and not be dependent on “chance” - another recurring theme in the book.

The second chapter, “Basic Philosophy Of Accident Prevention”, represents more or less the heart of the book, with 60 pages it’s also the second longest chapter. Heinrich begins by listing his 10 Axioms Of Industrial Safety. According to him these are ‘self-evident truths’ which are explained in the course of the chapter. Some of these most of us will subscribe to; others will probably be rejected at first sight, so they’re probably not all that self-evident… Or at least not anymore.

The first axiom discusses the completed accident sequence, usually pictured by the famous domino sequence which we find for the first time on page 13. What struck me is that Heinrich did come up with the metaphor, but apparently not developed it to the point how we currently use it. Heinrich has very much focus on the human, the person at risk, the one bound to make a mistake and get hurt. Also in this early depiction of the accident dominos this tendency to look at the person comes through very clearly. Heinrich’s sequence is (from right to left): Injury - Accident - Unsafe Act - Fault of Person and Social Environment, the latter dealing very much with ancestry and accident proneness. That’s a rather far shot from the current sequence of (for example) Consequences - Accident - Direct Causes - Underlying Causes and Lack Of Control/Management Factors.
While Heinrich is pretty much focused on the person and thus often gets cited in connection with behaviour based safety, he doesn’t neglect the fact that good accident prevention means to remove dangerous situations (i.e. mainly technical measures) and another recurring theme is his focus on management (notably first line management and/or supervisors). On page 23 he even argues that while machines in themselves don’t make mistakes, preventive measures on them are more effective.

Pages 16 and 17 present another surprise (and a quite shocking one, with the benefit of 80 years hindsight) when Heinrich argues that one has to focus on the direct causes. In his view the unsafe act of a person is the most important in the action sequence and removing that will prevent the accident from happening. Heinrich does hint on underlying causes (he lists examples like training and supervision) but doesn’t call them that way yet (he will do this many pages further on).
This focus on direct causes is obviously quite in contrary to what contemporary safety science teaches: focus on the underlying causes is eminent; removing direct causes is little more than cleaning up the symptoms. And Heinrich seems to have understood this in a way, since he says on page 20 that both kinds of causes should be eliminated as far as possible” (he talks here about human and technical causes, by the way), the example discussed on page 37 points towards a very diverse approach of accident prevention, and example 2 on page 38 is an excellent case of changing the environment that leads to changing the behaviour of persons. Still his continuous stress on unsafe acts and direct cause have certainly contributed to a great deal of misunderstanding of his work and led many a safety advisor and manager astray.

This misunderstanding is made even worse by the next theme Heinrich takes up: his (in)famous statement that 88% of all accidents are caused by unsafe acts. Of all the things Heinrich has written this bit is probably the item that has been taken out of context the most and twisted worst. If one takes just a little effort (really not much) one sees that 1) Heinrich talks about DIRECT causes here and 2) how he has found the 88%, which is rather arbitrary and not that much scientific at all. I have almost the feeling that Heinrich kind of fell in his own trap: he had the ambition to do something scientific and wanted to come up with a number, even if the way there wasn’t that scientific at all…

Page 27 pictures the other famous Heinrich metaphor: the triangle, pyramid or iceberg with a ratio between severe accidents and less serious incidents. He doesn’t call it this way, but he poses here the Common Cause Hypothesis (CCH), by stating that for every accident with consequences there are many other similar accidents without any consequences. This is another piece of work by Heinrich that has been bent, twisted and misunderstood in the course of the past decades. One, or maybe the, keyword in Heinrich’s statement is the fact that he talks about SIMILAR accidents. By keeping that in mind one will hardly proclaim that reducing minor incidents like slips, trips and falls will also reduce the number of major accidents like refinery explosions (see the Baker Panel report, the CSB Texas City investigation and everything on Deep Water Horizon).

It’s a sad thing that Heinrich’s brilliant idea of the CCH has been abused to so great extent that there are now even people who reject it all together - again without studying the initial proposal of the theory. And it has a lot going for it, doesn’t it. Heinrich makes a strong case by pointing out that the greater volume of minor incidents present an advantage to do something before a serious similar accident occurs. To quote the man (p. 30): “present day accident prevention is misdirected when it is based largely upon the analysis of major accidents” and (p. 32) “no accident, whether or not it results in an injury, is too insignificant to receive consideration…”. This was written more that 70 years ago and is still true today!

Heinrich does, by the way, a good job of separating consequences and the accident itself. For him it’s the potentiality that counts, i.e. he has focus on the risk. Oddly ‘risk’ is a word he hardly uses, if ever.

In connection with the pyramid one can say a few words on the 1 : 29 : 300 ratio that has become known as Heinrich’s Law among some. While Heinrich gets back to the ratio several times, he doesn’t appear to see this as a law himself. Even more, on page 30 he says that the number of no-injury accidents probably never will be known exactly! (By the way, on page 50 he has even a 1 : 82 : 2553 ratio…)

I talked already about causes. Heinrich is pretty much obsessed by those, even though a bit too much (for my and modern safety science’s taste) with direct causes. Another of his central messages is that causal analysis is necessary for the identification of countermeasures and accident prevention. On page 39/40 he for the first time poses the phenomenon of underlying reasons. A bit of a shame that he doesn’t expand his thoughts and theories on these.

In the final pages of chapter 2, Heinrich discusses a number of interesting thoughts that are highly relevant today:
- Safety and quality problems have the same causes and require the same actions (e.g. p.40).
- Safety is a management responsibility (p.43) with a special role for the foreman (see also appendix III).
- Accident prevention is good business (p.50, later also p.395) not at least because of the “hidden costs” for which Heinrich estimates a 1 : 4 ratio. On page 68 there is a strong paragraph that argues that safe factories tend also to be the more productive ones.
- The responsibility for the management is in Heinrich’s opinion rather a moral (p.43) or financial (p.50) one than that he says something about compliance (which appears to be the main driver for many a safety advisor these days).

Quite a lot of stuff for just one chapter…

Chapter 3 deals with creating and maintaining interest in safety. It contains a great deal of pseudo-psychology, but also contains a couple of sensible things (either supported by accepted theory or common sense), like the fact that one should think before doing and that it’s important to know your target group before acting. There are, however, also a few truly embarrassing and totally political incorrect and today unacceptable statements (like on immigrants, on p.75). Heinrich does mention a few useful angles for commitment raising campaigns, including (professional) pride as an important motivator.

The fourth chapter deals with Fact Finding. First Heinrich discusses the various causes: direct or proximate (acts and conditions), subcauses (personal factors) and underlying causes (managerial, supervision, social factors and environment). At this point I wonder why he never returned to his dominos and re-designed them with this distinction in mind. Others (e.g. Bird) would do this several years later… While Heinrich does mention underlying causes he regrettably chooses not to discuss these to any greater extend in his book. A missed opportunity, and as said before, a major source of misunderstandings afterwards and it weakens Heinrich’s work in retrospect because he keeps focusing on unsafe acts and direct causes (and measures to remove these). Even worse, in the next chapter (p.152) he even decreases the importance of underlying causes. A very strange argumentation.
On page 125 and 126 Heinrich poses a very strong statement for causal analysis as an essential element in improvement that is still valid today, and he says that if your causal analysis is good enough, you basically just need to “reverse” this cause to identify the remedy. At the end of the chapter he mentions that underlying causes may be the same for a larger number of different incidents. Without knowing it he hints here at the phenomenon of generic failure types and basic risk factors quite a few years ahead of Tripod and the like…

Chapter 5 deals with corrective action and accountability (especially for first line management) but leaves no further strong impression wasn’t it for one brilliant quote: “practicability and common sense must prevail in safety as in other things”. Sir, yes sir.

Chapter six is the book’s most extensive (nearly 100 pages) and for the greater part mainly of historical interest, dealing with the safeguarding of machinery and examples how to do this. The principles may still be valid today, but the machines discussed are out of use for many decades, I presume. A couple of interesting thoughts, however, also in this chapter:
- A nuance on his man-failure focus on p.168: it’s an important direct cause, but the environment has to be in good condition.
- Thus, the greatest safety gain and optimization is to be made in the design stage (p.169 and p.223/224).
- He introduces kind of a strategy for safety with taking on the source of danger as the first point of attack (p.170), this he will return to in the next chapter called Process and Procedure Revision with the central question: “why not do the job in another and better way”, again combining safety with efficiency and commercial gain.
- And, in a way he hints on RAMS on p.259: permanency, accessibility for repair, etc.

In the eighth chapter Heinrich takes on what he calls ‘safety psychology’. His definition of the word ‘psychology’ may clash with the understanding of many a current reader (including yours truly) and I can only say that Heinrich’s thoughts here are at least outdated. Various factors that Heinrich sees as reasons for accident proneness (and thus as psychological factors here) are definitely not (or at least not directly) psychology related: blood pressure, age, experience, skill.
The statement (p.269) that the safety rules should have “…enforcement no matter why not compliant…” is extremely dangerous and wrong. Of course it is of eminent importance to know why people are not compliant with safety procedures, how else can you help them to improve?!

Chapter 9 deals with fatigue. Most striking, and probably way ahead of his time, is here the observation that a machine should not dictate the tempo of the work. Also the remaining chapters are of lesser interest from today’s point of view, or rather specialist items like Chapter 11 on illumination. Heinrich does take on Occupational Diseases (10), First Aid (12), Accident Statistics (13, with too much space dedicated to a certain classification method, but good focus on registration of minor events - see p.344) and Education of Employees (14). The latter does include some interesting points, such as that safety education should be specific enough to make it relevant to the individual employee as well as another interesting quote (in my opinion against dumb compliance and for good leadership): “men go further when properly led than when driven”.

Chapter 15 contains a short summary in which Heinrich concludes that “action is the key”. Sounds good to me.

The book concludes with a bunch of appendices of which the first two are interesting from a historical point of view how safety has developed over the years. Other appendices do expand on a certain subject addressed in the main text, e.g. appendix III that discusses the role and responsibility of foremen and especially the urge to take that responsibility. Another slightly interesting one is appendix VIII that correlates age and accident injuries, showing that older/experienced workers have less injuries than younger colleagues, but that consequences on average are more severe. As far as I know an observation that is still valid (judging from some basic statistics during my years at the Haarlem workshop).

Details about the book:
Industrial Accident Prevention, H.W. Heinrich, 1941 (2nd edition), McGraw-Hill Book Company

To the next part of this article.

Discussing Heinrich, Part 2: Heinrich/Petersen/Roos '80

This is the fifth edition of the book and the first done without direct involvement of Heinrich, since it was done and published nearly 10 years after Heinrich had died. The book is heavily reworked from the second edition that I had read before, but it still includes some few passages that are copied literally from older versions. And to be honest, it’s awkward to have a 40 or 50 years old text between the other, relatively modern, texts (even though the readers who are unfamiliar with previous versions may not notice it - I did and I found it strange).

Now I haven’t read the 1950 and 1959 versions of the book, but it appears that Petersen (who seems to be the main writer/editor here) and Roos take a departure from some of Heinrich’s original thoughts and while keeping the title and some of the structure and thoughts as the basis they primarily aim to “look at how the principles have stood up to these exacting tests” and add a lot of newer knowledge along the way. One of the author’s aims is to give safety professionals a good professional knowledge, because “the person who has a sound knowledge of the principles… is well equipped to deal constructively with… all exposures” (something that sounds as an argument for generalist safety expertise).

Let’s have a look at the contents. This edition is split up in four parts. Part I covers the basis and philosophy of accident prevention. Part II deals with (a?) accident prevention method. Part III takes on some special subjects and Part IV contains the appendices.

The first chapter lays down the basic philosophy by introducing the author’s idea of accident prevention which includes both prevention and correction. While worker failure still is a central theme, there is attention for both an immediate approach and a long term approach, including controlling acts and environment and working on things like competence. Also Pedersen & co see the same improvement principles for safety and production.

A number of rather simple models is introduced, including Kepner Tregoe (p.9) and MORT (p.10/11) and we are presented with the statement that “errors are inevitable” - compare that to earlier work of Heinrich. This is, by the way, the first of the 6 criteria of MORT. Good stuff. The MORT performance cycle is basically a (less complete) variation on the famous Plan - Do - Check - Act of Shewart/Deming. Page 12 presents a combination of various models. What I like in this model is the inclusion of principles and beliefs in the most basic level - so they do include safety culture, only they didn’t know that concept at the time yet, I guess.

Chapter 2 is basically a critical evaluation of Heinrich’s 10 axioms. And as we will see, safety science has in the meantime come a longer way than the original axioms which are heavily revised at places. The most striking example of this is probably the updated domino sequence presented on p.25. The one we see here, is the commonly current day used one by Frank E. Bird jr. including Basic Causes and Lack Of Control as the most right dominos instead of Fault of Person and Social Environment as Heinrich originally proposed. It’s also Bird who is quoted at length and discusses critically the (former) focus on direct causes.

The book presents several other variations on the dominos, including one by Edward Adams (p.30). He has an odd (IMHO wrong) understanding of the terms operational and tactical (the wrong way around) but worse in his version is the forced way in which he tries to plot organizational levels into the domino sequence. His detailed descriptions of elements in the dominos, however, are valuable and the fact that he includes “personality of the organization” (again, culture without calling it that) at the most basic causal level is very good. Quote: “Management structure is a reflection of the beliefs, objectives and standards of the key decision makers of the organization”.

Weaver’s variation on the dominos (p.31) is even less successful. He again includes terms like “Fault of person” and “Heredity”, pointing at the person and the question about “whether” suggests a lot towards counterfactual reasoning.

P.33 presents a model by Zabetakis that gives an interesting perspective. Strong element is that Zabetakis more or less equates the direct cause of an accident with an unplanned release of energy; a good thought and useful shortcut in many occasions. Disadvantages are that the model is a bit too complex to keep easily in your mind and the connection between management factors and other causes is not clearly described, although the model does include plenty of management factors. Another model is the Stair Step Model (p.38/39). The idea of using tolerance boundaries appeals a lot to me, but regrettably the model itself is too complex to explain and therefore nearly useless in practice.

One further expansion on Heinrich’s work is Multiple Causation. Heinrich’s original understanding was fairly linear, but p.36 takes on critically this “narrow interpretation”. And justly so; the dominos, linear and singular as the metaphor may be, need not be a limitation to multi-causation. Also root causes are discussed here.

The chapter continues with a number of other models, including the Motivation/Reward model (p.44) that may be useful to explain behaviour (even though there are even better ones, like the Theory of Reasoned Action of Ajzen & Fishbein, but then this was published in the late 1970s and may not yet have reached Petersen…). The Petersen accident model (p.49) looks a bit multi-domino-ish. Rather strong feature is the human part here regarding the time of writing - James Reason’s seminal work still had to come, mind you. Pretty weak, however, is the sidestep with ‘System Failures’ none of which can be the direct cause of an accident, but the diagram seems to suggest this.
Another theoretically sound model appears to be the Firenze model. Disadvantage is that the graphic representation is near impossible to explain. An interesting thought here is that “accident causation involves consideration of as many variables that affect the system as possible…” something that hints a bit of FRAM, many years later…
The Ball model (p.54) states that “all accidents are caused by hazards” something which I disagree with. A hazard is a necessary condition for an accident, but since a hazard also is a consequence of something desired (traveling by train with high speed, having a drink on your balcony, hot coffee) it should rather be regarded as part of the context than as a cause.
The final model presented is the Surrey model (p.54/55) which is very handy and logic, but alas cursed with an (IMHO) unpractical graphic representation.

The authors conclude this lengthy discussion of various models with a piece of sheer wisdom (if you ask me): “we believe there is no single best model”. Amen to that!

Section 3 of the second chapter takes on the subject about humans and machines. There is a very interesting discussion on the discrepancy of “humans cause” and countermeasures aimed at machines. A really intelligent observation, and well handled. There’s also a discussion of how Heinrich came to his 88% of human causes. Regrettably after this rather intelligent introduction there is a (for me) rather stunning twist: the authors lock in on human failure and even criticize the OSHA law that addresses mainly conditions (I would think, by the way, that the law takes on hazards and countermeasures, not as much causes). A very strange passage in this book.

The fourth section discusses the pyramid. The book states that the 300 : 29 : 1 ratio originally was viewed as an opportunity, and I think that’s an excellent point of departure. After this we’ll get the Bird update of the pyramid (1 : 10 : 30 : 600) and a discussion of the confusion about the ratios and a sensible conclusion is that “it does not mean that these ratios apply to all situations…”, “…the triangle for electricity is different from the one for handling material…” and “…common sense dictates totally different relationships in different types of work” (all p.64). All of which is very much in line with my thoughts about the Common Cause Hypothesis (which doesn’t get called this way in this book either). I think, however, that Petersen et. al. screw up a bit with their understanding of the phrase ‘frequency’, addressing only minor injury and not near misses and conditions/acts that would fit in the same CCH-group.

Most of the other Heinrich axioms come out rather well from the discussion and update. Petersen and his team agree with the responsibility for management and say even rather boldly: “In safety we’ve gotten into the ludicrous position of pleading for management support instead of advising how management can better direct the safety effort to attain its specified goals”. Keywords: communication, responsibility, authority and accountability! Bird appears to have a different view on the role of the supervisor than Heinrich. In Bird’s opinion the supervisor does exactly what the big bosses want, so according to Bird the key persons are not the direct managers, but the highest up in the organization. Weaver, in turn, does focus on the supervisors just like Heinrich.

The theme about hidden costs is another that gets some discussion and revision. The 4 : 1 ratio is dismissed with “that was in 1926” and some newer studies are mentioned, especially the Bird iceberg (p.89) and wisely the authors remark: “to actually quantify hidden cost is an almost impossible task and probably not worth the effort. If management believes in the concept it is often unnecessary to quantify”.

The chapter ends with the safety-efficiency axion that has a lot going for it (unsafety leads to loss, i.e. it is not efficient), but with ETTO (something nobody had written about in 1980) in the back of your mind also against it.

Having discussed the basics of the original Heinrich books, we enter Part II of the book on ‘Accident-Prevention Method’, or shouldn’t it rather be ‘methods’ instead? Chapter 3 is basically about two flowcharts (well, kind of) of the accident handling process from fact-finding and analysis to measures and monitoring of those. The various steps from these flowcharts are discussed in further detail in chapters 4 to 9.

Chapter 4 deals with collecting and analyzing data. Slightly unlucky is the fact that they talk about a subcause”; while they earlier in the book discussed multiple causes, it’s a bit unclear if they propagate linear single causal connections after all. What I like a lot is that they do (yet without saying it explicitly) discuss the relative character of the terms “cause” and “effect”. Really good, since people tend to forget that every effect (or consequence, if you want) can be the cause of something else. And what is absolutely a good thing of this 5th edition in comparison with its predecessors, is that it breaks with Heinrich’s focus on direct causes by (justly) stating that these are merely symptoms of underlying weaknesses.

Pages 109-111 present Weaver’s TOR tracing system for root causes. A high ‘investigation by checklist’ level here, so not my thing. And then the ‘whether’ question… I mentioned already the danger for counterfactual reasoning. Another problem I have with this is that the authors (and Heinrich before them, by the way) entirely forget to discuss risk assessment as the (or at least a) basis for safety management. Also do the authors tend towards a compliance driven ‘whether’ later in the chapter. Not their finest moments here.

Chapter 5 discusses System Safety and finally we get some discussion of hazard identification, even though I do not fully agree on the author’s definition of hazard. But that’s sifting through details. What I do like a lot is their thoughts on the occurrence of accidents through either 1) energy transformation (though unsafe act or condition) or 2) energy deficiency (which causes a safety critical function to fail, e.g. a plane’s motor). Handy and often applicable.

Accident Analysis is the subject of the next chapter. Take care that the authors apply a certain meaning to the words ‘investigation’ and ‘analysis’. Many of us may think about ‘investigation’ in terms of fact finding and checking hypothesis about the causes while ‘analysis’ is about the identification of the various causes. When the authors talk about accident analysis they mean the analysis of trends. It is very interesting to see how important the authors think that trend analysis is, but then, they saw the iceberg/pyramid as a major opportunity, so it shouldn’t be surprising to read that they find this a “necessary prerequisite to the selection of an effective remedy” (p.134 and also bottom p.135). I fully agree.

Quite funny from today’s point of view: a lot of fuzz about how easy working with a computer can be. Duh!

Chapter 7 deals with the selection of a remedy. For starters they discuss the ‘old Heinrich axiom’ and conclude with that the “engineer the hazard out” as first choice is just a good and sound advice. After this follows a whole bunch useful stuff about cost/benefit analysis, Kepner Tregoe’s 7 steps and a bit of Fine and Kinney. The following chapter, Application of Remedy, is for the greater part a direct copy of older versions of the book and then ends up in a lot of details, e.g. about slips and falls.

Chapter 9 deals with Monitoring, and it’s good that the book has a lot of attention for this important issue. As the authors say, reasons for monitoring could or should be: 1) gauging performance, ‘how are we doing’, 2) providing us with red flags when thing go wrong, and 3) a measure of performance for the responsible individuals. A major part of the chapter (around p.210) is quite useful as input for an eventual self-evaluation questionnaire. As for a methodology of monitoring an important tip is provided: “the past performance of any group is the best standard to use as a guide for present performance” (p.225) because people in the group understand it and will accept it, and obviously it will measure if there is improvement. On the same page the authors even discuss ‘leading’ indicators - excellent! A shame that they also dwell on LTIF for a while… And one good point that underlines Bird’s opinion on managers: “managers are first and foremost motivated by their bosses’ measures of their performance” (p.226).

Chapter 10, Creating and Maintaining Interest, is for the first part again a direct copy of Heinrich’s original text. After this we get a quick run through of a lot of behavioral theory on motivation (Maslow, Argyris, Herzberg), learning (Skinner), interpersonal skills, communication and management theory. As a whole this is a great improvement on Heinrich’s original pseudo-psychology and a fine summary of some important psychological theory for the safety professional. Basically recommended reading.
One more worthwhile quote: “The key to employee safety performance is a management that cares and does something to demonstrate that caring. The key to supervisory performance is doing something to make it matter to the supervisor whether or not he performs in safety”.

Chapter 11 picks up the Motivation Model from chapter 2 once more and in more detail. First we get another short discussion of accident proneness (which was already dismissed in chapter 2 in a most effective way by applying statistics and normal distribution - on some other grounds too. Very strong piece of applied science if you ask me… on p.42): “accident proneness is real, but extremely rare”, so with regard to accident prone susceptible people: “don’t waste time on that either” (p.261/262). And on we go to more useful matters like ability determinants, proper placement (where I would have said something about guidance and tutoring, but maybe that’s included in training and supervision?), training, climate (“the way we see the company safety program strongly influences the behaviour and ability to learn”, p.262), if it’s a lively company (lively being a good characteristic of the company’s culture and learning ability) and the friendliness of safety people (an interesting point that may serve as a mirror for some safety professionals).

A large part of the chapter is actually used on discussing the safety culture/climate in a company. Page 264 addresses the discrepancy between climate perceived by the workers and the climate that is intended and poses the question if enough effort has been spent guiding the philosophy down the line. If the superior does not reflect the organization’s philosophy there will be a perception discrepancy! Again ahead of their time (Schein’s work came after this book I believe - have to check) the authors concluded that safety climate (culture) is a product of both the corporate climate and a functioning safety program (p.268).
The text goes on listing Scanlon’s basic climate requirements: 1) Overall goals and objectives; 2) Objectives communicated; 3) Goals distributed over departments; 4) See the big picture instead of sub-optimalization; 5) Meaningful participation of the individual and 6) Freedom to work, relatively autonomous. A good list to keep in the back of your mind. Page 265 has an interesting discussion of communication and the danger of ending up in bureaucracy. One key element in a good safety climate is a turn from authority orientation (do as you’re told) to goal orientation (“if a worker explains the job to do/done and what it will achieve for the company”). It’s also argued that stability is essential for a good climate, which is typical because stability and the mentioned expansion aren’t necessarily simultaneous things, but expansion and job security maybe a bit more…
Likert promotes participative management because this leads to higher levels of achievement (p.266), something that is worked out further in the text, e.g. by supportive relationships, group decision making and group methods of supervision. Participation is essential according to the authors: “participation and its resultant commitment become a substitute for authority”.
The chapter concludes with a discussion of what motivates and what not. Quite a surprising list of satisfiers (things that lay in the job itself) and dissatisfiers (stuff peripheral to the job, at best functioning as “short time teasers”, like payment - even though people do adopt behaviour and perform actions for positive reward).

Chapter 12 deals with Safety Training and once more copies a big deal of original text from Heinrich. Most striking is the conclusion that the method of training isn’t as important as other variables including motivation through working towards a goal, reinforcement, practice, feedback, meaningfulness and the climate/culture. And: “if the supervisors are thoroughly sold on safety, they in turn will sell the employees whose work they direct” (p.277). To support this then chapter 13 has a ‘formula for supervision’. The text discusses the paradox in supervisors being key personnel for quality, safety, maintenance and direct control of employees, but that this is hardly ever taught to them. The text does not give a 100% solution, but if I can summarize it comes very close to the famous plan-do-check-act principle. And a very wise observation: supervisors don’t have the same motivators like ordinary employees, exactly because of their position.

The last few chapters are of lesser interest, dealing for a major part with insurance issues and legislation that is utter irrelevant for professionals outside the USA and probably superseded by now for those in the States. Most interesting is Chapter 16 on Risk Management which also makes clear that risk management and safety management don’t always have the same goals. Sharpest division is when they discuss the function of the safety manager (who aims to control/prevent losses) and a risk manager (who also funds for losses that occur, and will even try to transfer the risk to someone else, the latter being hardly acceptable in safety).

This book also concludes with a number of appendices of which again the historical one are of greatest interest.

Concluding, I would say that this is a very useful update of Heinrich’s original work with a good critical approach and a lot of new knowledge added. Much of the book (especially the central parts) stands to this day and it would have nice to have seen a sixth edition with updated knowledge from the past 30 years of safety science. To my knowledge no such book exists, however. What does exist is a very critical book on Heinrich which we will take on next.

Details about the book:
Industrial Accident Prevention, H.W. Heinrich, Dan Petersen, Nestor Roos, 1980 (5th edition), McGraw-Hill Book Company (ISBN 0-07-028061-4)

To the next part of this article.

Discussing Heinrich, Part 3: Manuele

In comparison to the other two books, this one is strikingly thin (87 pages including everything), only A5 sized with an enormous font and something like double line space. As you can imagine something for a good afternoon’s read and no more.

Manuele’s aim was to eliminate “from our practice any old but still used premises that, upon examination, would be judged invalid” (p.1). A very worthy cause in my opinion, because there’s little I hate more than (so-called) professionals selling bullshit.
Since a lot of questionable premises are found in Heinrich’s work, Manuele set out to “review the origin of those premises, how they have changed over time, and their validity” (p.1) the results of which ended up in this booklet.

Manuele discusses the four original versions of Heinrich’s book among which there are apparently significant changes, something I was unable to check since I only have one version, but it doesn’t surprise me - the four versions came out over a 30 year period with approximately 10 years in-between each update; it would have been strange if there had been no changes. Sciences and the world move on, after all. What baffles me, however, is that Manuele doesn’t make one single mention of the 1980 reworked edition, especially since several (maybe even all?) of his comments are addressed and corrected in that version. No doubt that Manuele must have been aware of this fifth edition; he even quotes other work by Petersen. So I wonder why Manuele does neglect the 1980 version. Maybe because it makes his own booklet rather obsolete? Anyway, let’s see what Manuele has to say (all of which is neatly summed up in Chapter 11 that can nearly in its full been read on Heinrich’s Wikipedia page).

The introduction tells us a bit about Heinrich’s life and work and then lists the outline for further discussion. What surprises a bit is that Manuele excuses Heinrich for his out-of-date and sometimes sexist style of writing ("consider the times in which he wrote” - p.5), but doesn’t use similar considerations with regard to the difference in safety science we have at our disposition these days and the little that Heinrich had way back then.

Chapter 2 reflects on transitions in the work world from the 1930s to 2000 and then poses the question if workplace studies made in the 1920s still be valid today. This question is left rather unanswered, expect for an “I don’t think so”. Manuele uses a bit of statistics first, but as far as I can judge he throws together stuff you can’t just compare without further work (after all, disabling injuries and lost time injuries are NOT the same thing). Further more I’d like to pose the hypothesis that while a lot has changed in the past 80 or 90 years, the basic principles can be very valid today. And probably are.

Chapter 3 discusses psychology and safety and what Heinrich meant about this. As I said in my discussion of the ’41 book this certainly wasn’t Heinrich’s finest moment and much of what Manuele says I just agree with, although Manuele just like Heinrich doesn’t define what he understands as ‘psychology’ thus weakening his own point.

Chapter 4 takes on Heinrich’s 88 : 10 : 2 ratio in (direct) causes. Manuele says that he believes “that those who proclaim that unsafe acts are the principle causes of accidents do the world a disservice” (p.19). Amen to that. Fully and unashamed. Manuele thinks that this ratio has done the greatest harm to the practice of safety “Because when basing safety efforts on the premise that man failure causes the most accidents, the preventive efforts are directed at the worker rather than on the operating system in which the work is done” (p.22). I would say that the ratio coupled with Heinrich’s focus on direct causes has done the greatest damage. The ratio in itself, dealing with direct causes after all, could very well be a right one depending on what and how you count and what your definitions are. The problem, however, is that this may (will) be utterly insignificant for real substantial safety work since that should concentrate on underlying and root causes instead. And that focus is something that Heinrich does address (and suggests) at times but at other moments totally neglects or even downplays.

Chapter 5 deals with the pyramid and the 300 : 29 : 1 ratio. I wonder why Manuele is so obsessed by the exact ratios? We are safety professionals and not accountants, aren’t we? But no, he finds it necessary to discuss the numbers (something that Heinrich himself was not so straight about, and something that gets a lot of nuance in the fifth edition). It’s almost as if Manuele doesn’t (want to) see the big picture, greater “opportunity” (see the discussion of the 1980 edition) and the beauty of the metaphor. Also he totally neglects the Common Cause Hypothesis as the central thing. Quite disappointing for someone with so many safety science merits as Manuele apparently has. The next chapter carries on with the theme and I think that had Manuele taken the CCH as the basis in mind, this entire chapter would have looked differently. He now only focuses on consequences and goes on arguing that the causal factors between serious accidents and minor incidents are totally different. Of course, since he leaves out one crucial thing: they have to be similar accidents. The CCH, remember?! Just looking at the frequency of consequence is bullshit of course. And here Manuele goes astray just as Petersen et.al. in their update.

The 4 : 1 ratio of hidden costs is the subject of Chapter 7. Yes, Manuele does like his numbers. And maybe this is because it’s such an easy area to prove Heinrich wrong (which has been done in 1980 already, by the way, but in a much more forgiving way). Wouldn’t it be wiser instead to conclude that, okay, the numbers are not applicable always, but the big picture (the actual loss is much greater than just the damage/injury) is right?

Chapters 8 and 9 take on Heinrich’s Principles of Accident Prevention and Axioms of Industrial Safety, but it escapes my understanding what point Manuele is trying to make here. I feel that he agrees with some, disagrees with others but he isn’t really clear why and how.

Chapter 10 discusses the accident sequence (dominos) and makes some of the observations I have done with regard to the left dominos. Regrettably without tying in the more recent updates from the likes of Bird. I really don’t think that anybody actually uses the original Heinrich dominos anymore and he is only credited for coming up with the original metaphor back in 1931 while actually everybody uses one of the revised versions.

Details about the book:
Heinrich Revisited: Truisms Or Myths, Fred A. Manuele, 2002, National Safety Council/NSC Press (ISBN 0-87912-245-5)

To the next part of this article.

Discussing Heinrich, Part 4: Reflection and Conclusion

I took the liberty to insert quite some of my own thoughts and remarks in the ‘reviews’ of the three books before, so I’ll keep this short for now.

After having spent ample time with Heinrich’s work, I’d like to credit him for the groundbreaking work he has done. Sure, I have read none of his contemporaries (if there are any) but in some things he was ahead of his time, just as in some other things his writings exactly reflect the spirit of the times. But in that respect he no doubt is like any other author. And like any piece of science, it should be taken, tested, refined and rejected, something we have seen that has been done. So even if we strip a lot of out-dated stuff from his work, he has left us some worthwhile principles (e.g. the Common Cause Hypothesis) and at least two brilliant metaphors (the dominos and the pyramid) that are easy to communicate even to safety novices.

What disappoints me with Heinrich’s work is that he in a way wasn’t consequent. It’s as if he hasn’t thought through things far enough. Why did he get stuck in his focus on direct causes? Why did he have this focus on unsafe acts? He mentions the concept of underlying causes, and he discusses at length other measures than human-directed ones, yet doesn’t make this his key message. Was it his insurance background, or maybe a deeply rooted belief and misunderstanding of psychological factors? Or did Heinrich work on the premises that if the environment was as it should, the last thing remaining was focus for the human acts? He doesn’t say this explicitly and anyway it wouldn’t entirely be so. Whatever it is, I don’t know the answer to these questions. But luckily we have the benefit of hindsight and 80 years of additional safety science to pick the valuable bits out of his work and do a god job.

As for Manuele… in The Netherlands there’s a saying: “De beste stuurlui staan aan wal”, meaning that those who think they know best and have most critical remarks don’t actually participate in solving the problem. That pretty much sums up much of my impression of Manuele’s booklet. He makes the job for himself very easy by attacking mostly things that are easy to ‘falsify’ (notably the numbers) but not coming with any sensible suggestions for improvement. Most of Manuele’s chapters leave me with a very unfinished feeling: he disproves something, but not quite: he often doesn’t finish the job and leaves arguments hanging in the air. Which also applies to critical remarks he makes underway, for example addressed to the entire Behaviour Based Safety movement. He just doesn’t work out his statement into something concrete.

And then the question why Manuele chooses to neglect the revised 1980 version of Heinrich’s book as well as all the updates of Heinrich’s models by people like Bird (who improved both the dominos and the pyramid and also the ‘hidden cost’ principle)… Highly questionable if you ask me.

At any rate - one of the most important skills of safety professionals is to be critical and question things all the time. Including their own scientific (or something ideological) basis. So for me this was a worthwhile exercise. Now it’s your turn!

Frist comment on Heinrich on EHSQ Elite

Before writing my discussion on Heinrich and Manuele there was a thread on the EHSQ Elite group on Linkedin to which I contributed.


Here's the post that triggered me:

What concerns me is SHE's continued belief in Heinrich. I would encourage all reading this thread to read "Heinrich Revisited - Truism or Myth" by Fred Manuele, published in 2002.

Heinrich's work is truly myth, including the triangle. Colleagues, please take time to read Mr. Manuele's work and then comment. The beliefs in Heinrich's myths keep us focused on less serious but frequent incidents believing that reducing their risk will somehow reduce risk of serious and fatal injuries where the exposures may be completeley different

My reaction:

Thanks for the tip on the “Heinrich Revisited” book. I wasn’t aware that it existed so I ordered it right away. I still have a copy of Heinrich’s 1941 book waiting to be read, so that’ll be two in one go… Meanwhile I checked out the Heinrich article on Wikipedia and there one can find what looks to be a good summary of Manuele’s criticism. I found some good and valid points here, but also some that I’d like to attribute to wrong understanding and the like. This activated me to sit down for some writing since I’d like to raise some critical comments myself…

The first would be that it’s rather easy to just sit there and criticize Heinrich’s work. Why not rather take the effort to update/upgrade his work to today’s standards and for example write a book “Heinrich Revised” instead? As I say often to managers and employees: It’s really easy to complain, but much more useful to take that extra step and try to improve.

It seems to be fashionable these days among some safety experts to do a bit of Heinrich bashing. What annoys me a great deal is that people tend to take Heinrich’s findings literally as if it were laws of nature (e.g. the 1-10-30-600 ratio, or whatever the distribution is, see the reference in Wikipedia to “Heinrich’s Law” - ridiculous) and then criticize Heinrich upon finding out that it isn’t. Or even worse: people posing conclusions that their own registrations aren’t good enough because they don’t live up to the “correct” ratio (I don’t make this up - happened at a forum I attended - twice within one hour!). What bullocks is that?!

Heinrich’s work at the time was plainly groundbreaking. I don’t recall anyone before him approaching safety in a semi-scientific way and leaving us at least two concepts/metaphors/models (the dominos and the pyramid) that are still somehow relevant and applicable today. Yes, certainly his work does have shortcomings. But don’t forget that we’re 80 years further down the road, have a truckload of additional safety science in our luggage and have had ample possibility to test, try, refine and partly reject his findings. Which is what a major part of scientific work is all about.

What is inexcusable, in my opinion, is when things get taken out of context and are rejected based on that wrong understanding or twisted application.
(To tackle that other Heinrich-thing in that regard: are there still people who believe that, in analogy with the brilliant and easy to explain/understand domino metaphor, all accidents are a simple and linear chain of events? Ever heard about the concept of lies-to-children?).

Let’s discuss the ratio a bit further. I still have to read it in Heinrich’s own words, but as I interpret this in the line of today’s safety science: it is totally unimportant whether the ratio is 1-10-30-600 or rather 1-25-60-500 - which probably would make a rather odd pyramid, haven’t tried to draw this. I can tell from my own experience (having the luxury of working in companies with over 15.000 to 30.000 registrations of all kinds of incidents per year and thus ample data to play with) that each incident type may have specific and probably due to circumstances changing ratios (Andrew Hale wrote at least one paper on this subject in the late 1990s/early 2000s) which may result in distributions that aren’t very pyramid shaped at all. Linda Bellamy mentioned a diamond shaped one; I would like to add an example from my own line of work. When we look at collisions of trains with track workers you have an incident type that would generate a kind of uneven hourglass. Collisions with trains are pretty binary events: either the subject dies or it’s a near-miss (rather “near hit”). There are only few ‘happy’ people hit by a train that are in need of a first aid kit or consult with a medic…

The point to be made with the pyramid is that GENERALLY (and luckily) there are only few fatal accidents, more with injuries, even more with material damage and even more near misses. Heinrich’s premises was thus that attacking the frequent low-consequence incidents will eventually reduce or prevent the number of serious accidents as well. Seems to make sense, if only because it’s pretty tedious to wait for the really strong signals to appear - much better to (re)act on weak signal and prevent the strong ones from happening at all.

Here then are at least two of the misunderstandings that often occur:

1) People tend to forget what it was that Heinrich actually counted. This (in)famous pyramid consists of data from a huge number of various incidents (wasn’t it something like 1,5 million?) which he ranged in a certain way according to the consequences resulting in this pyramid-like distribution. Meaning several things: Firstly Heinrich not only counted apples and pears; no he went out to get the entire fruit market from coconuts and bananas to raspberries and tomatoes and ranged these from rotten to ripe to green summing up these categories. The pyramid thus collects things that are very, very different in nature. Secondly let’s get another thing very clear: the pyramid is about CONSEQUENCES and consequences alone. Which gets us to point 2.

2) Often forgotten and/or misunderstood is one of the central things in Heinrich’s work, namely the Common Cause Hypothesis. This says that serious accidents and its precursors (i.e. minor incidents) are due to the same causes. Thus providing a theoretical grounding why one should do something about the “weaker signals” to prevent the serious accidents from happening. Crucial here is of course the COMMON CAUSE. Logic dictates that we will find a huge number of different causal chains (or rather tree - see the domino comment) within our fruit market (*) and looking at preventive measures that work one has to remain within one causal tree and not blend these together. It’s folly to think that preventive measures on apples will do something about the rotten coconuts (replace “apples” by “finger cuts in kitchen” and “rotten coconuts” by “BLEVE” if you like). Yet this is the un-nuanced message that is often spread by ignorant/lazy colleagues and consultants. It’s hardly Heinrich’s fault that this happens, or is it? (We could draw a parallel to religions and fundamentalism here by the way, I won’t now).
(*) Note: And likely a couple of underlying “causes” that cover a much larger selection, but these are usually so general/generic/non-specific that they are rather classification terms (think of catch-all phrases like “competence” and “culture” or the GFT/BRF from Tripod) than real causes in the chain.

Anyway, the case is: while part of Heinrich’s conclusions may not be literally valid anymore (if they ever were, literally), the general line is applicable today as much as it were then. Zooming in on a certain accident/incident type and its precursors (instead of taking a consequence as the point of departure) both a pyramid(ish) ratio (specific to each type and probably company) can be found and the common cause hypothesis can be applied. And even better: it has been scientifically tested. Sure, it took a long time since 1931, but read for example Linda Wright’s PhD (2003) and I guess the Storybuilder project in the Netherlands can provide additional evidence here.

Concluding with my own view on the title of this thread: wishing for zero fatalities is obviously a good thing but as a goal it’s useless. Just how are you going to steer on such infrequent events as fatalities or LTIs (see various eminent authors on this subject; apart from the fact that you can only go so far with controlling external factors - at least in my business). And it may do actual damage as some of our group members justly argued. In our company we rather talk about a zero-vision, not goal. Goals are to be set on the weaker signals that may lead to an accident (and eventually fatality), or rather on some pro-active leading indicators. But that was another thread, wasn’t it?