Elsevier

Safety Science

Volume 29, Issue 3, August 1998, Pages 163-177
Safety Science

Article
Safety rules: procedural freedom or action constraint?

https://doi.org/10.1016/S0925-7535(98)00020-4Get rights and content

Abstract

The paper presents a partial classification of safety rules as constraints imposed from outside on the freedom of choice of individuals or companies. The classification is linked to the Reason model of skill-, rule-, and knowledge-based behaviour. The imposition of constraints is related to a number of criteria concerning the context and use of the rules to arrive at proposals for when and where to apply what type of rule. To illustrate the principles discussed an example is given of the rule from the Dutch Working Environment legislation concerning the requirement to use “current state of the art in prevention” in companies. The paper proposes providing procedural rule support to companies on how to meet the rule, in the form of a data bank of practicable solutions to health and safety problems.

Introduction

`Safety rules' must be one of the most widely used terms in the field of safety science. It is also one of the most misused, largely because it is rarely carefully defined before the author or speaker goes on to draw conclusions about whether, and in what circumstances, rules are a good or a bad thing.

A random selection of requirements which have been called safety rules in one context or another illustrates the range:

  • 1.

    “Wet paint: do not touch.”

  • 2.

    “Safety helmets will be worn everywhere on this site outside office buildings.”

  • 3.

    “Before welding is carried out in confined spaces a check will be made of the adequacy of ventilation.”

  • 4.

    “All employers will ensure that a risk assessment of the company operations is carried out by a competent expert and will be repeated when significant alterations to the operations take place.”

  • 5.

    “The level of exposure to risk outside the company's perimeter shall not exceed 1 in 10−6 chance of death per year.”

A definition of safety rules which would cover all of these examples might be as follows:

A safety rule is a defined state of a system, or a defined way of behaving in response to a predicted situation, established before the event and imposed upon and/or accepted by those operating in the system as a way of improving safety or achieving a required level of safety.

Such a definition indicates that rules are designed to limit the freedom of choice in a given situation by imposing defined ways of responding. The definition further leaves open the type of system state or behaviour which is the subject of the rule: the way it is generated, imposed or accepted, who is expected to comply with it (e.g. individual or company) and the way in which the improvement or level of safety are defined. Each variable in the definition can be classified into a number of sub-groups, often on several different dimensions, producing a very large matrix of types of rule. We must expect that many statements about safety rules will only be applicable within a sub-set of the boxes in that matrix. Until there is some measure of agreement about the matrix and what fits where in it, it is predictable that much discussion will produce more heat than light. The first part of this article attempts to provide some of the building bricks for the matrix and to locate some of the discussions about safety rules within the suggested edifice. The second uses the structure to discuss one specific rule and how it could be applied; the application of the `known state of technology' to the design of workplaces.

Given the definition of rules given above, the first question to ask is: what falls outside its scope? The answer is: very little. All the rules and habits which a person has in their head for coping with routine and previously encountered situations are covered by it. The only situations which fall outside are those where neither the individual nor company concerned, nor others in the system who have some right to specify behaviour, have predicted or met the situation and none of them have defined appropriate behaviour for it, but have accepted that the response to such `unexpected' situations should be spontaneous and without prior planning. Not surprisingly, this corresponds to the category of knowledge-based behaviour in the Skill–Rule–Knowledge (SRK) classification (Reason, 1990), which is generally regarded as the least reliable and predictably safe mode of behaviour. Even where an individual has acted at the knowledge-based level, but the authorities consider that the behaviour should have been rule-based, a safety rule can be considered to have been violated.

The message of cognitive psychology has been that the vast majority of individual behaviour is to a great extent preprogrammed. People learn and become efficient by developing and storing their own rules and applying them to appropriate situations they meet. The judgement of `appropriateness' is one of matching rules to patterns in the situation. The message of quality management is very similar (though rather more normative); that the behaviour of employees and managers in effective, high quality systems follows rules and that the vast majority can and should be planned in advance.

If much of behaviour follows rules, then safety rules are merely that sub-set that improves or achieves safety. With such a definition safety rules can never be a `bad thing', unless we accept that too much safety can be a bad thing.1 To locate the element of controversy which we have said exists around safety rules, we need to scrap the words in my original definition (above) “... and/or accepted by ...”.2 We then end up with the following definition:

A safety rule is a defined state of a system or way of behaving in response to a predicted situation, established before the event and imposed upon those operating in the system, by themselves or others, as a way of improving safety or achieving a required level of safety.

This localises the discussion about safety rules to the issue of their imposition and underlines the fact that the discussion about safety rules is not about whether there should be safety rules, but who should develop or specify them. In the end some individual has to decide what concrete action to take in any given situation and has to live with the consequences of that action. The question to be decided is when is it best to leave the persons taking the action to decide for themselves how to proceed safely (either in advance=rule-based behaviour, or at the time =knowledge-based) and when it is appropriate and/or acceptable that their freedom of choice be limited by getting someone else to decide what the action should be. The limitation of freedom can be partial or complete, depending on how the imposed rule is formulated.

If we turn Reason's SRK hierarchy on its head, we can use it to define the stages in the formulation of rules; by their nature they have to be established at a knowledge-level of problem solving, be formulated as procedures for establishing a rule and then specified in detail for (routine) execution. This gives three main types of rule which also link to the three levels of the hierarchy of safety management systems (Hale et al., 1997)—goal setting and evaluation, policy development and procedures, and execution of work instructions (see also Hale, 1990, for a more detailed discussion of a classification of safety rules related to the SRK classification). The types of rule also strongly influence and are influenced by, the method of establishing whether someone has complied with them, so we need to consider this in each case.

  • 1.

    1. Rules defining goals to be achieved; for example:

    • 1.1.

      maximum acceptable concentration of toxic substances in air or water;

    • 1.2.

      defined risk contours of fatal accident frequency rate around a factory site;

    • 1.3.

      “the highest possible level of safety” (Article 3 of the Dutch Working Environment Act); and

    • 1.4.

      “as low as is reasonably practicable” (ALARA).

These impose no rules as to how the goals should be achieved, but simply imply that sanctions will be imposed if they are not met. Their practicability and the way in which they are regarded by the authorities and those who must carry them out depend to a great extent upon the measurability of the goal (see also further below and in part 2 of the article). If the result cannot be seen or measured easily, then the only way to assess if someone has obeyed the rule is to check what they have done and how they decided to do it. This transforms the rule effectively into one of type 2 or even 3 below.

  • 1.

    2. Rules defining the way in which decisions about a course of action must be arrived at (how and by whom), (= solution search rules); for example:

    • 1.1.

      “taking into consideration the known state of technology” (also taken from Article 3 of the Dutch Working Environment Act);

    • 1.2.

      “where standard operating procedures cannot be used the course of action to be taken will be determined jointly by the control room supervisor and the maintenance supervisor”;

    • 1.3.

      “the employer shall consult with his employees or their representatives in establishing his safety care system”; and

    • 1.4.

      defined sets of release and dispersion models for calculating risk contours.

Such procedural rules for searching for the solution to the safety problems which arise may define the people who must be involved in the decision, the methods to be used for searching, the parameters which must be considered in judging the appropriateness of the course of action, etc. The detailed course of action is still left fairly open. Assessing whether the rule has been followed is a matter of checking whether the procedure has been followed and evaluating the quality of the decisions made at each step.

  • 1.

    3. Rules defining concrete action or required states of the system:

    • 1.1.

      “Safety goggles will be worn at all times in the laboratory.”

    • 1.2.

      “All lifting tackle will be inspected at least once a year by a competent authority and defective equipment will be withdrawn from service and destroyed or repaired.”

    • 1.3.

      “Each pressure vessel shall have two independent pressure relief systems.”

    • 1.4.

      “The medical adviser in the expert service shall have passed an approved examination and shall be inscribed in the professional register ...” and

    • 1.5.

      “No smoking.”

These rules remove almost all freedom of choice,3 though there may be room for interpretation in the wording, which may be gratefully seized upon by those required to conform to the rule in order to enlarge their scope for discretion (when is defective not defective? what is approved? etc.). Checking compliance is a simple matter of observing behaviour or measuring the state of the system, the presence and functioning of required hardware, etc.

A number of the advantages and disadvantages of imposing more limiting `action/state' safety rules can be summarised as follows (see also Elling, 1991):

Broadly speaking action/state rules, as opposed to goal rules, maximise the advantages, but also the disadvantages.

In summarising these pros and cons we have tried to avoid any specification of who the rule imposers and followers are. The implication is that, whether the rule follower is an individual inspecting his own workplace, a designer designing a new machine, or a company setting up its safety management system, there will be the same repression of innovation and resentment at loss of freedom if detailed action rules are imposed from outside. The company director of a large company dislikes detailed laws which restrict his business freedom as much as the skilled worker dislikes petty rules about how to exercise his craft. But equally the director of the small company who has no time to devise his own safety rules will be just as pleased as the novice worker who is only interested in next week's wages when someone else does their thinking for them and defines rules and laws in a concrete way.

One possible way out of the dilemma of resentment at imposed rules is if the rule follower and the rule imposer are the same person, or at least that the latter consults the former in order to involve him/her in the rule making. The justification for and implications of such a combination are outside the scope of this paper (e.g. Hale, 1990, Elling, 1991). It is also not always possible or affordable, because of the difficulty or cost of consultation, e.g. for national legal rules for transport safety or even occupational safety. The rest of the paper confines itself to situations where rules are imposed on one party by another.

A problem with this classification is that, although the distinction between the three levels of rule is theoretically clear cut, it may be blurred in practice because of problems of interpretation and measurement. A number of the goals given as examples cannot be measured directly and have to be redefined in terms of procedures for measurement or criteria for assessment; otherwise it is not possible to agree whether the rule has been obeyed or not. It is this issue of monitoring of rule following and assessment of rule compliance that, more than anything else, complicates discussion of safety rules. The more that compliance is stressed, the more rules are used as a basis for assessing (rewarding and punishing) behaviour, and the more that there is dispute between parties over the behaviour expressed in the rules, the more there is pressure to define the rule at a management or regulatory level in terms that are easily measurable and verifiable. This reduces the complication and expense of disputes of interpretation. In the more litigious countries (e.g. Anglo-Saxon) this has tended to favour action/state rules, which are often, in their turn, felt then to be unjustifiably restraining by those at a lower level who have to carry them out. Action/state rules are particularly prevalent in the safety area where there are no easily definable output goals, in contrast to occupational hygiene, where more easily measured goals in the form of maximum allowable concentrations can be defined.

Our thesis here is that we should see the rule as a progressive limitation of freedom of choice; limiting first goals, then how decisions are arrived at and finally limiting the actions themselves. The key question is how we decide who is best placed in what types of situation to impose what level of limitation. At what level in the hierarchy of an organisation, industry, country, etc., should the step from type 1 to type 2 to type 3 rules be taken. These levels range from an individual at one end to a supranational organisation (e.g. EC, ILO, UN) at the other. It is usually convenient to distinguish the individual and small (work) group levels (micro), from company or industry levels (meso), and those in turn from national or supranational (macro). It is also relevant to ask what type of people, e.g. workers, managers, experts, manufacturers, consumers, etc., take the decisions at each level on behalf of themselves or others.

It is impossible in one paper to work out this question for such a broad range of situations and groups involved. What we shall therefore do is to offer some general statements about the factors and principles relevant to the choice and then move on to a practical example.

At least the following factors are relevant (see also Hale, 1990):

(A) How predictable and closed-ended is the system? The more predictable, the more the action/state rules can be worked out and agreed once and for all at a high level. This implies establishing them at least at a meso-, if not macro-level. The more exceptions there are due to local circumstances or differences in technology, the nearer the action/state rule-making needs to be to the micro-level.

(B) How rapidly does innovation in the system (need to) take place? The more rapidly, the lower the level at which the translation to action/state rules needs to be.

(C) How much interaction is there between individuals at the micro-level in the system, whereby it is important that they follow the same rules? The more this is so, and the more the interacting individuals are spread over different physical regions or different organisations, the higher the level at which the action/state rules must be agreed and imposed.4 The harmonisation of road traffic rules, signs and infrastructure design is a case in point.

(D) How likely are people to pursue their own goals within the system, to the detriment of the system goals or of others whom their actions affect? The more this is so, the more pressure there is to impose action/state rules from the level above that at which the conflict expresses itself. This is related to the question of checking for uniform compliance (see above).

(E) How much knowledge is there at the micro-level about how to make decisions? The greater the professionalism and experience at that level, the more rule-making can be left there.

With these general principles in hand, and with the overall (libertarian) philosophy that higher levels should so far as possible avoid imposing restrictions of freedom on lower levels we now proceed to discuss a practical rule-making case.

Section snippets

Solution data banks and `the state of technology'

The situation chosen is the introduction of an Article in the Dutch Working Environment legislation (Arbeidsomstandighedenwet, 1980) phrased as a general requirement (at the goal level) to keep safety policy and safety measures up to date, in line with advances in the state of knowledge about hazards and preventive measures. This requirement has been introduced into the legislation of many countries as a recognition of the fact that safety is not a static science. New technologies make it

Discussion and conclusion

Both of these answers to the question of what are generally known measures (standards and solutions data bases) have been formulated in terms of a procedural rule governing how to search for what is known.

The first answer can be formulated as the following rule: “If there is a C-standard (action/state rule) follow that. If there is no C-standard or you do not want to follow it, use Fig. 1 as a procedure for analysing the hazards and coming to your own rules from first principles.”

The second

References (17)

  • A.R. Hale

    Safety rules OK? Possibilities and limitations in behavioural safety strategies

    Journal of Occupational Accidents

    (1990)
  • Arbeidsomstandighedenwet, 1980. (Working Environment Act 1980) SDU. Den...
  • CEN, 1991. Safety of Machinery: Basic Concepts, General Principles for Design. Part 1: Basic Terminology, Methodology....
  • Elling, M.G.M., 1991. Veiligheidsvoorschriften in de industrie (Safety rules in industry). Ph.D. thesis, Faculty of...
  • European Economic Community, 1985. A council resolution over the new approach to standardisation and harmonisation....
  • Hale, A.R., 1978. The role of HM inspectors of factories with particular reference to their training. Ph.D. thesis....
  • Hale, A.R., de Loor, M.H., Huppes, G., van Drimmelen, D., 1988. Algemeen erkende regelen der techniek en...
  • Hale, A.R., de Loor, M., van Drimmelen, D., Huppes, G., 1990. Safety standards, risk analysis and decision making on...
There are more references available in the full text version of this article.

Cited by (138)

View all citing articles on Scopus
View full text