Adonis Diaries

Article #13, April 10, 2005

 “How basic are task taxonomies in Human Factors?”

The follow up question is: how can we conceive practical human error taxonomies before working out taxonomies for the tasks required in a system, its processes or steps in a method? 

If the type of skills required by an operator to perform a set of tasks are not well defined and studied it might not be that useful to apply a complex error taxonomy that does not delineate the applicable domain. For example, how can we allocate functions to either operators or machines or how can we decided who is better at performing a set of tasks an automated machine or a trained operator if we fail or cannot classify the human capabilities and limitations versus the potential capabilities and limitations of the machine we intend to design?

There is a relationship between task taxonomy and task analysis.  Originally, task analysis methods were conceived to break down a job into work modules and then to elemental tasks that standard time measurements could be applied to in order to maximize profit on human efforts. The purpose of task analysis is to originate an ordered list of all the task that people will do in a system with details on information requirements, task times, operator actions, environmental conditions, evaluations, and decisions that must be made.

Consequently, a task analysis should produce estimates of time and effort required to perform tasks, determination of staffing, skills, and training requirements, pinpointing the necessary interfaces between operators and the system, and to provide inputs to reviews and specifications.  This process enables detailed examination in the evaluation of human functions in terms of abilities, skills, knowledge, and attitudes required for performance of any function from inputs to outputs.  When profits are the bottom line you should also have in mind that reducing errors is a major criterion beside time saved and direct costs.

It seems implicit when allocating standard times that the appropriate conditions of work are explicitly defined, the age and gender of the worker are acknowledged, the duration and frequency of rest breaks accounted for, the eventuality that overtime work is considered and ability to cope with boredom and repetitive tasks because all these variables would affect the standard times for accomplishing a job efficiently with minimum errors for the long haul.

If you were to decide between the two alternatives: either correcting standard times to finish a task based on experiments accounting for the above factors that might affect efficiency, safety and health of workers, or allocating a separate expense fund based on actuarial studies for rate of illnesses, rate of errors, hospitalization cost and overturn among workers if the uncorrected standard times are applied, then which choice would you definitely retain?

A task analysis of a system allow estimate of the likelihood of a certain error (i.e., the product of frequency and the probability of occurrence of a certain error) and how often the error will occur for a duration, thus enabling a numerical estimate for the acceptability level and need for a redesign.

The consequences for lack of a task analysis combined by practical error taxonomies in designing a system are not that futile on operators, end users and the whole performance of systems since time is of the essence for delivering a functional product. 

The fact that current technology can automate the travel of airplanes from take off, to cruising and to landing without the need of a pilot does not guarantee safety or acceptability by airplane commuters.

The obvious problem is who in his right mind would board an airplane without a certified pilot and a co-pilot? It seems that in Japan the fast trains have no train pilot aboard but are controlled before reaching destinations.  In this case, passengers are taking these trains but would rather be doubly secured by having trained pilots on board no matter the extremely high safety records of these automated trains.

Nowadays, most of these functions and task allocations are done by computer programs with the hope that an expert professional is going to take serious time to analyze the printouts and provide a judicious human feedback. These computer programs have, crossing our fingers, the necessary constraints on safety standards, health standards, serious errors restrictions and labor requirements for the least.

A student provided a version of the “Shel” model as a standard task taxonomy that would permit sharing of data among different modes of transportation and other industries.  Apparently, this model can serve as an organizational tool for data collection in the investigation of workplace.  The components of the Shel model are 1) Live ware (the individual to human interface); 2) Hardware (human to machine interface); 3) Software (human to system interface); and 4) Environment (human to environment interface).  The model might relate all peripheral elements to central human live ware and thus focus on the factors which influence human performance.

The best way to assimilate the concept of task taxonomy is by examples.  For the purpose, one of the assignments is to study the job of the bread earner of the family, through questions, observation, and investigation and analyze its task taxonomy. Another assignment is a lecture project analyzing the task taxonomy of an industry or system not covered in the course materials.

Are you wondering what methods could be used in Industrial engineering, Human Factors or Industrial Psychology for improving designs?  Would you be interested at working taxonomy for methods in the next article?