One death is a tragedy and a million deaths is a statistic
Most lectures at occupational health and safety conferences or zoom webinars invariably include death by Pierrepoint presentations from vainglorious behavioural scientists or Pecksniffian safety crusaders, which are often embellished using classical statistics. It typically reflects a positivist worldview with an inordinate emphasis on scientism and an excessive dependence on ex post facto, statistically unreliable and enigmatic performance metrics. These include lost time or serious injuries, medical treatment cases and total recordable injury frequency rates, which are used to fulfil the sophism of what gets measured gets done. Frequency rates are inexplicably linked to lucrative performance bonuses and it inadvertently ensures that what gets measured is manipulated and the metric gets managed instead of safety performance.
Presentations are littered with references to statistical reports and dashboards with bewildering infographics from regulatory authorities, statutory agencies and various quangos. Much of the discourse is pseudointellectual cowardice and fails to challenge the orthodoxy. An indoctrinated audience is often captivated or beguiled by the big data, triangles, elaborate graphs and colourful pie charts. The worthless statistics are typically and foolishly considered much better than no data at all………Though it’s cold and lonely in the deep dark night, I can see paradise by the dashboard light.
At most safety conferences or webinars, cohorts of narcissists bask in the micturating patronage and relish dropping obscure statistics, irrelevant complex terminology and other non-sequiturs into discussions, which masquerades the paucity of any formal education. Even some cursory exploratory data analysis confirms most presentations and reference materials are littered with fundamental errors. The insipid and indoctrinating language is peppered with turgid sesquipedalian corporate hogwash, which further corroborates that most safety crusaders are often economical with the truth or thicker than a wharfie’s sandwich and usually unfamiliar with their portfolio……You took the words right out of my mouth.
Exploratory data analysis is an integral component in the daily work of many professions and in an era placing enormous faith in big data, remaining reliably informed requires some basic knowledge of statistics. This includes descriptive and inferential statistics and many other fundamental terms covering the classification, interpretation and analysis of data. Initial considerations consist of several elementary steps and requires establishing the number and type of variables present and determining the specific measurement scale used for each variable. It enables adept practitioners to organise data into meaningful displays and deliver coherent, consistent and reliable messages to enhance decision making.
Most safety proselytes cannot distinguish descriptive from inferential statistics or explain the difference between qualitative and quantitative data or confirm if the control and response variables are being measured or counted. The prolific and sacrosanct risk matrix is usually categorised as a quantitative tool although its numbers are merely labels and it often enigmatically underpins informed decisions regarding material or operational risks in most corporate risk management programs and supplementary risk registers. It is reminiscent of phrenology and no better than a Hasbro Ouija Board.
Glazed eyes with furrowed brows soon appear if any safety evangelists and many behavioural scientists are asked to describe number scales and explain the difference between ordinal, nominal or categorical data and establish the mean, median or mode as an estimation or gauge of central tendency. Basic statistical notation such as discrete and continuous variables, which include an independent control variable and dependent response variable are guaranteed to induce perplexity. This is antagonised by elementary questions relating to sample size, accuracy or validity of specific test scales and their interval or ratio scales of measurement.
Many safety crusaders embrace oversimplified, linear and mechanistic cause and effect relationships and are often bamboozled by the statistical caveat that correlation does not imply causation. The bewilderment increases exponentially with inferential statistics, which enable us to estimate characteristics of a defined population and use past processes and supplementary performance data to make informed decisions regarding the future. This is based on the principles of probability using confidence intervals and probability or ρ-values. It can be accomplished via suitable hypothesis testing, which involves framing the null (H0) and alternative (H1) hypotheses, choosing an appropriate test statistic and critical value, establishing the level of significance (α-level) and accepting the null hypothesis (H0) or rejecting it in favour of the alternative hypothesis (H1).
The null hypothesis (H0) represents the status quo or establishes the situation remains the same and any observed differences are due to chance. The alternative hypothesis (H1) traditionally denotes what you are trying to prove. It can only be accepted if the probability associated with the null hypothesis (H0) is extremely low and very unlikely to be attributed to chance.
The University of Ballarat offered a graduate diploma in Occupational Hazard Management (ED4), which included a module entitled Statistical Methods in HSE (MG410) under the astute guidance of Peter Martin and Jack Harvey. An assignment at its Mount Helen campus involved an arbitrary selection of academic transcripts from various scientific journals. The intake was divided into randomly selected groups and each team was required to deliver an appraisal of their allocated research paper, which included analysis of descriptive and inferential statistics and hypothesis testing protocols. The results and conclusions from each presentation were quite astonishing and remarkably consistent.
Most of the selected research papers provided reasonable summaries of descriptive statistics although the inferential statistics, especially hypothesis testing protocols were littered with fundamental errors. Several of the research papers used inapposite hypothesis test methods and the framing of null (H0) and alternative (H1) hypotheses was often tenuous. In many cases the probability or ρ-values, test statistics, critical values and the level of significance (α-level) were unidentified and never disclosed. The selected papers were published and retrieved from various notable academic scientific journals but would barely fulfil criteria covering a Cochrane systematic review.
This is further complicated by reduced federal or state government funding to tertiary education establishments. It has forced numerous sandstone and redbrick universities, business schools and colleges to seek support from many renowned corporate brigands and the curriculum reflects the objectives of its platinum corporate sponsors, which inevitably align with the Friedman doctrine. It distorts the quality, diversity and standard of education, which eventually deteriorates and generates a cacophony of indoctrinated parrots holding superficial degrees or doctorates.
This unhealthy alliance with corporate sponsors also has a detrimental effect on research activities. Any funding can compromise independence and emasculate the integrity of epidemiological health studies or surveys. Its extent is quite sinister and scientism has an established tradition of manipulating evidence, data and analysis to create and maintain an auspicious environment for industry at material and ideological levels, which is especially prevalent throughout the pharmaceutical sector.
Meanwhile, the industrial safety sector in Australia and its peak safety body continues with its crusade of mindlessness in an otiose attempt to shoe horn reality into a positivist architecture of oppression. This is underpinned by a stem worldview that places an unwarranted emphasis on rational decision making and operant conditioning via antediluvian behaviourism. It is reflected via an anachronistic safety curriculum with a traditional hard and soft systems management paradigm supported by a prebaked body of knowledge, which is inapposite to an environment dominated by increasing volatility, complexity, ambiguity and uncertainty. Its proselytes are victims of an ideology or arrogance and invariably do more talking than listening and fail to recognise the Dunning-Kruger effect and accept that on almost every subject someone else knows much more than they do.
Our law courts are often required to reach decisions under conditions of radical uncertainty and the positivist acolytes of probabilistic reasoning, Bayesian statistics and Poisson distributions have much to learn from the accumulated wisdom of the legal profession and common law. Court proceedings are the most systematic process available for reaching decisions in situations entangled with imprecise knowledge or tenuous cause effect relationships. This often involves some historical ignorance with even greater apprehension about the future and it is best addressed via abductive prudence, which is remarkably different from the probabilistic reasoning of classical statistics.
In many cases there is insufficient information or evidence to frame the situation in probabilistic terms using a credible and coherent declaration and any civilised judicial system must always treat its citizens as individuals and not as a discriminatory statistical distribution unit. The doctrine covering presumption of innocence implies there is no Bayesian dial above a judge’s head in a court of law. Its primary objective is to establish the context via descriptive inquiry, which precedes probabilistic reasoning. Bare statistical evidence without sufficient narrative is never enough and justice must be administered in a distinctive or individualistic manner and not via a prescribed mathematical formula or equation.
Quantitative risk analysis includes parameters such as probability, frequency and exposure but without a credible and coherent explanation, it is very unlikely to secure a conviction in a court of law. Abductive reasoning involves a quest for the best explanation or a persuasive narrative that offers perspective and context relevant to the specific case. It provides beliefs and values with coherence and significance and may even elucidate or corroborate evidence.
Over many centuries narrative reasoning has embodied tales of a nation’s development. The life of law has been experience, which underpins legal decisions, not logic. It cannot be summarised by the axioms and corollaries from a mathematics text book or annual statistical reports embellished by regulatory authorities and statutory agencies, which merely prove that science advances with each funeral.
Although judges will direct jurors to reach a verdict based solely upon the available evidence, articulate narrative reasoning plants the seed by establishing the context. This is crucial to the rationale and operation of trial by jury and is the initial step in the risk management framework and it asks the question…..What is going on here?
An eloquent and righteous explanation demonstrates internal coherence and social sensemaking, which enables a judge and jury to understand the event in its entirety and recognise exogenous and endogenous risk factors. It provides integrity and substance and its consistency reflects and aligns with most of the available evidence and general knowledge that is readily accessible to the judiciary and jurors. Statistical reasoning has some pertinence but it must be integrated into the entire narrative or best explanation. Superior narratives are often unique, discernible and easily distinguished from incoherent explanations littered with complex probabilistic reasoning although maybe stories are just data with a soul but……No one ever made a decision because of a number, they require a story.
Statistical reasoning can assist in the process of rational deliberation but should never replace or even precede narrative or abductive reasoning. In an era of increasing volatility, complexity, ambiguity and radical uncertainty, the world reacts to unique events and is governed by individuals not statistical distributions. In each individual case, the progress of truth is not necessarily the same as the dispensation of justice and it requires a process of legal or abductive reasoning, which respects distinctiveness, peculiarity and ingenuity. A courtroom is a place where stories are told, evaluated and judged but it need not be the only place.
Much like statistical or probabilistic reasoning, rational decision making is underpinned by positivism and it dominates the safety curricula, the OHS Body of Knowledge and the black box psychology of behavioural safety. It disregards the enigmatic interdependency of our 3.14159 recurring subjective minds in an otiose attempt to square the circle via one objective brain and technocratic rationality.
This positivist stem worldview endorses a binary paradigm using a false rational and irrational dichotomy, which simplifies and often distorts judgement and disregards the crucial enigmatic aspect of arational decisions. Human decision making is incongruous with deposits and withdrawals from an accounts ledger or the inputs and outputs of an information processing system and it never resembles approaching an and/or gate in an event or fault tree analysis diagram using Boolean algebra.
Rational decision making disregards the phenomenology of perception and many other concepts such as intercorporeality, interaffectivity, satisficing, heuristics and semiotics. The brain merely hosts conversations and does not issue commands or act independently from its body, which is verified by a substantial corpus of evidence. This is reinforced by the dynamic power of the collective unconscious that is well recognised amongst megabillion dollar multimedia advertising agencies although it is often summarily dismissed in the risk and safety sector in favour of probabilistic reasoning via the universally sanctified risk matrix.
Much of this is exacerbated by a relentless quest for perfection using business improvement techniques such as Six Sigma or Operational Excellence and the brutal fatwa of zero harm, which masquerades corporate turpitude under a rubric of righteousness. It denies fallibility and renunciates wisdom and the following lyrics from Jim Steinman resonate:
You’ll never find your gold on a sandy beach
You’ll never drill for oil on a city street
I know you’re looking for a ruby in a mountain of rocks
But there ain’t no Coup de Ville hidin’ at the bottom of a Cracker Jack Box
Meanwhile, Australia’s peak safety body is holding a virtual rock concert from its Tullamarine glasshouse amidst the Covid-1984 pandemic. It features a splenetic crusade against the federal government Infection Control Expert Group. Indeed, evidence suggests this panel of distinguished professionals and other infection control groups…..do not yet fully understand the problem.
Evidence from intensive care units at many major public and private hospitals across Australia indicates significant noncompliance with even the most basic workplace health and safety requirements and the following mandatory warnings signs and other controls are required in most hospitals:
· Danger!!!!! Door Opens Inwards!!!!!!!!!!
· Danger!!!!! Detergent – Do Not Drink!!!!!!!!!!!!!
· Danger!!!! Very Hot Water!!!!!!!!!!!!!!
· MAINTAIN THREE POINTS OF CONTACT ON STAIRWELLS!!!!!!!!!!!!!
SAFETY IS YOUR RESPONSIBILITY!!!!!!!! stickers must be strategically located on the back of toilet doors and washroom mirrors and the ZERO HARM logo should be embroidered into work uniforms and operating theatre gowns.
Behavioural safety observations are required on operating theatre surgeons, which must focus on fundamental unsafe acts and target situational awareness, eyes on path and eyes on task. This is endured by many electricians and other qualified tradespeople across constructions sites every single day. It is in our DNA and embedded in our culture, which is………the way we do things around here.
Our peak safety body and many of its proselytes are quite willing to assist and control the monster they have created under its rubric of righteousness. The primary object of statutory safety legislation is to secure the health, safety and welfare of people at work although citizens also have a right to expose charlatans, fraud and deception………Baby we could talk all night but that ain’t getting us nowhere.
Over to you safety thought leaders and delusional bloggers.
The author acknowledges the esteemed work of Peter Martin and Robyn Pierce, especially their joint publication entitled Statistics for Starters (Six Sigma Publishing, Buninyong 3357, Australia – 2002), which was used as reference material covering the statistics section in this article.
Bernard Corden says
……”Scientism has an established tradition of manipulating evidence, data and analysis to create and maintain an auspicious environment for industry at material and ideological levels.”
The following link to the Grenfell Tower inquiry supports the comment above:
An inordinate emphasis on positivism eventually entangles the system and its processes with many of the impurities it attempts to resolve.
This was quite evident following the DNRME response to the Queensland parliamentary inquiry into black lung and its doublespeak “in principle” agreement to the recommendations.
Meanwhile, the current mine dust lung diseases toll in Queensland nudges towards 200 victims:
How come black lung all but disappears in the Hunter Valley and Illawarra when you drive across the border into NSW?
Then have a read of Bad Science and Bad Pharma by Ben Goldacre:
………”There are lies, damned lies and statistics” – Mark Twain