In the 19th century, French doctor François Magendie questioned the effectiveness of blood-letting in treating fever. He had a fierce disagreement with another clinician, Dr. Broussais, which led to accusations and counteraccusations. Magendie eventually encountered what he believed were falsified statistics suggesting that blood-letting reduced fever mortality. He reacted with disgust and outrage, saying:

"Thus, the alteration of truth, which already manifests itself in the progressive form of lies and perjury, offers us, in the superlative degree, statistics."

Mark Twain later popularized the saying “lies, damned lies, and statistics.” Today, while blood-letting is no longer a medical practice, statistics are still often used to paint an overly optimistic picture of uncertain realities. In safety management, we still rely heavily on statistics. While regulations necessitate their use, there is often a disconnect between these statistics and the actual management of worker risk.

Until the early 1900s, workers who suffered occupational accidents bore most of the burden, along with their families. Employers' liability insurance did not cover workers and often relieved employers of moral responsibility. The advent of Workers’ Compensation Systems improved compensation for injured workers. In some parts of the world, these systems allocate compensation based on injury rates.

A major problem with injury rates is that they can rationalize and normalize work injuries instead of preventing them. Policymakers often favor lower compensation premiums. Ongoing adjustments to compensation benefits are mediated by insurance boards influenced by various social and economic factors. Companies can pay settlements and move on without addressing the underlying causes of injuries. Thus, Workers' Compensation Systems based on Lost Time Injury (LTI) rates can facilitate the underreporting of workplace hazards and misrepresentation of workplace realities.

From the early years, there have been efforts to transform safety from a cost into a profit-generating aspect. The introduction of LTI rates did not end incidents; these rates are unsuitable for making decisions about risk management. Furthermore, the rates are susceptible to manipulation, such as assigning administrative tasks to injured workers or excluding self-employed workers from statistics.

Having no accidents does not necessarily equate to a safe work environment. Therefore, insurance premiums and company actions regarding health and safety risks should not solely be based on the occurrence of accidents—especially not on a rate that can be manipulated—but on proactive, preventative, and mitigative measures that address accident mechanisms and their underlying systemic factors. To safely manage system performance as a dynamic process, we require a variety of data. Because we cannot foresee everything that happens within the system, rather than chasing statistics, I advocate for the use of qualitative data from the system.

Sources:

Eastman, C. (1909), The American Way of Distributing Industrial Accident Losses: A Criticism, in: American Economic Association Quarterly, Apr., 1909, 3rd Series, Vol. 10, No. 1, Papers and Discussions of the Twenty-First Annual Meeting. Atlantic City, N.J., December 28-31, 1908 (Apr., 1909), pp. 119-134.

Finkelstein, R. (2022), Lost- Time Injury Rates: A Marxist Critique of Workers’ Compensation Systems, in: New Scholarship in Political Economy, Vol. 216/17, Leiden: Brill.

White, C. (1964), Unkind Cuts at Statisticians, in: The American Statistician Vol. 18; Iss. 5.