Human errors in operator-machine systems

Accidents at work often stem from interactions between people and machines, where mistakes can lead to harm or damage. These human errors occur when operators interact with equipment in ways that deviate from safe procedures. Understanding these errors helps in designing better safety measures.

Types of human errors

Human errors can be categorised based on the nature of the mistake made during machine operation. This classification highlights common pitfalls that contribute to workplace incidents.

System errors in operator-machine systems

Beyond human mistakes, accidents can result from flaws in the design of machine systems themselves. System errors occur when the interface between the operator and the machine is poorly designed, making it hard to use safely. This includes issues with displays and controls, which are key components of operator-machine interactions.

Displays

Displays provide information to operators through visual or audio means, helping them monitor and respond to machine status. Visual displays include lights, dials, and digital readouts showing data like temperature, speed, or time. Audio displays use pre-recorded messages or tones, which are useful in low-light environments or for mobile tasks, and are often employed for warnings or time-sensitive alerts.

Ways to reduce display errors:

• Match the display type to its purpose – digital displays offer precise readings, while dials suit quick glances.
• Group related displays logically, such as placing all indicators for one system together.
• Position displays near their corresponding controls for easy reference.
• Use colour coding to highlight information and signal warnings.

Controls

Controls are the mechanisms operators use to interact with machines, including levers, switches, handles, joysticks, pedals, and keyboards.

Ways to reduce control errors:

• Group controls logically, based on the sequence of use and frequency of use.
• Ensure controls are within easy reach and simple to operate.
• Add protections against accidental use.
• Label each control clearly.

By addressing these design flaws, workplaces can minimise system-related errors, creating safer environments where operators can work more effectively.

Reducing accidents using token economies

One effective way to decrease workplace accidents is through behavioural interventions rooted in psychological principles. A token economy is a system based on operant conditioning, a learning process where behaviours are shaped by consequences. In this approach, desired behaviours are reinforced with tokens (secondary reinforcers), which act as symbolic rewards that can later be exchanged for tangible benefits (primary reinforcers), such as goods or privileges.

Fox et al. (1987)

This study aimed to investigate introducing a token economy system in two USA open-cast mines where there had been accidents.

Method

• Workers from each mine were divided into hazard groups: Group 1 (office jobs) were least hazardous, and Group 4 (electricians, scrapers, and operators) were most hazardous.
• Employees earned tokens for working without time lost for injury, not being involved in accidental damage to equipment, and for behaviour that prevented accidents or injuries.
• Tokens were lost for unsafe behaviour.

Results

• There was a large decrease in days lost through injury as accidents were reduced.
• The system was used at one mine for 12 years and at the other for 11 years.

Conclusions

• Behavioural programmes are effective at reducing accidents in the long term.

Evaluation

• Strengths - The long-term data collection provided evidence of lasting impact, and the practical application saved lives and reduced costs.
• Limitations - The study was specific to US mining contexts, limiting generalisability to other industries or countries. Additionally, smaller accidents might have been underreported due to the incentive structure.

Monitoring accidents at work

To prevent accidents, workplaces must first monitor and record them systematically. This involves identifying patterns and causes, allowing for targeted interventions. Without consistent methods, many accidents go unnoticed, hindering prevention efforts.

Swat (1997)

This research focused on improving accident documentation in Polish industrial settings, where existing systems were inadequate. Poland's system of documenting accidents was ineffective. The aim was to develop an organisationally useful method of recording risk to find causes of accidents and prevent them.

Method

• The research method involved case studies and interviews with line managers, safety supervisors, and employees.
• The design was longitudinal.
• The research involved four industrial plants from different branches (foundry, machinery, meat processing, and furniture) in Lodz, Poland.
• All of the plants were old, with old equipment. The plants employed 2964 workers in 1993.

Procedure

• Accidents in 1993 were analysed for frequency, severity, and causes by accident reports, researchers' investigations of accident protocols, and interviews with safety supervisors and line managers.
• Minor incidents in 1994 not resulting in death, serious injury, or serious damage were analysed from the meat processing plant, using data on first aid cases and interviews with 96 employees.

Ethics

• Confidentiality was upheld, though if there were only one of a certain type of factory within that area, it may be easily identified.
• The results could be of benefit to the workers.

Results

• Accident frequency averaged 2.8 per 100 employees, with severity at 44.6 sick days per accident. The foundry had the highest frequency (5.9), while machinery had the highest severity (61.2).
• Common accident types included manual work (1.1 per 100 employees, highest in meat processing: 1.2) and working with machine parts (0.7, highest in foundry: 2.6).
• Four essential causes of accidents were identified: insufficient supervision (highest – 89%), poor workplace organisation, technical factors (lowest – 11%), and human error.
• Poor housekeeping (e.g., slippery floors, faulty staircases, improper tools, incorrect clothing) contributed to 65% in the foundry, 33.3% in machinery, 47.8% in meat processing, and 37.5% in furniture.
• In the meat processing plant (1994), there were 254 injuries requiring medical treatment and 23 cases with sick leave days. Employee interviews suggested total incidents could be as high as 520, meaning many were not reported.

Conclusions

• The type and location of accidents and incidents should be recorded.
• Poor maintenance/housekeeping should be noted as a key cause.

Evaluation

• Strengths - Data triangulation (combining reports, investigations, and interviews) increased validity. The longitudinal approach allowed detailed trend analysis.
• Limitations - Subjectivity in reporting could bias data, as employees might underreport to avoid issues. Generalisability is low, as the sample was specific to Polish plants and not representative of all industries or countries.

Issues and debates in health and safety research

Research on workplace health and safety often explores broader psychological themes, helping to understand why accidents happen and how to study them effectively.

Individual and situational explanations

Explanations for accidents can focus on individual factors (e.g., personal errors like omission or timing mistakes) or situational ones (e.g., poor system design or insufficient supervision). Swat (1997) used data from accident reports and individual interviews, taking into account both individual and situational explanations.

Idiographic versus nomothetic approaches

A nomothetic approach seeks general laws, as seen in Swat (1997), which analysed patterns across plants to identify common causes. However, a more idiographic approach with qualitative data collection would allow investigation of individual reasons behind accidents. Thematic analysis of responses could provide accident prevention information.