Adonis Diaries

Article 41

“What do Human Factors measure?”: An example of warning alarm systems

This article is an ongoing project.  The dependent variables or performance measurements adopted by Human Factors/Ergonomics are varied and should be judiciously selected to correspond to the tasks, systems, methods, and purposes they are researching.

The professionals in the field of human factors, depending on their primary discipline and interest, consider appropriate performance measurements for the types of controlled experiments, evaluation, or testing methods.  The studies  either answer practical problems that cannot be resolved by the traditional methods that analyze systems’ feasibility and performance or are oriented toward basic research.

The dependent variables should be necessarily directly related, or highly correlated with the essential human factors performance criteria, of mainly reducing errors in the operations, safe usage of products/systems, and health concerns of workers and end users.

Many of these measurements and their corresponding techniques and procedures were initially developed by psychologists, who were attuned to the practical facets of their discipline, and how their research can be applied to engineering design of interfaces between target users and systems that are increasing in complexity and becoming essential in running the functioning of our daily lives.

For example, I asked my class to participate in resolving a problem.  We were to evaluate two warning alarms used in factories and to decide which product to select.  I asked them what could be the potential, valid, and effective dependent variables for this evaluation.

Obviously, the first answer was to measure how loud the alarm is:  this variable would not do because alarms are designed to be loud in order to alert workers and employees, but the data on how many of them heard the alarm goes off will certainly not provide a decisive choice among the mostly loud systems. Loud is not a performance criteria in alarm systems but an engineering objective.

The next step was to make the students think about the real purpose of having a warning alarm.  Obviously, warning alarms are produced to warn against the existence of fumes, the starting of fire, a serious danger, or a drill.  Suppose you were conducting a drill, then what might you be interested to observe and record?

At this junction, ideas started to fuse from every corner;: the noise of the alarm should be jarring and very uncomfortable to the ears, visual stimuli should be designed in the product such as in police cars; automatic connections to fire brigades should be contemplated when installing warning alarms; more than one exit door should be opened; emergency exit doors should not be blocked by inventory materials.

Now, where the warning alarm should be installed?  Should it be next to the emergency door because this is the normal direction where heads turn to, or in the opposite direction because the reaction of people is to move in a direction away from the warning noise?

I meant to explain the fundamentals of evaluation methodology and ended up with practical answers, which is fine and encouraging, but we had to get back to methodology and what could be the most appropriate safety measure.

Most probably, how many workers actually vacated the premises on hearing the alarm could be a good starting decision variable.

Better, how many vacated within a standard duration in accordance with safety regulations might be an improved measuring stick.

At this stage there was confusion in discriminating among, controlled experimentation, evaluation, and testing methods, as well as the differences among dependent, independent and control variables.

Obviously, what we were discussing was testing the effectiveness of warning devices, since evaluation is targeted for the packaged deal that includes after sales services, maintenance, repair, instructions manuals, extra commendable features, and so forth.

In addition to rating each specification, evaluation methods might consider comparing end-users behaviors such as like/dislike, acceptance, or rejection depending on psychological preference judgments.

For example, we were supposed to be discussing how to test among warning products, but it was not straightforward to the students.  Students were talking of other factors such as the frequency range of the alarm, whether it is in the lower or upper frequency scale, the timber of the alarm, the loudness range and levels to manipulate, the characteristics of the background noise, the layout of the facility, the type of noises emanating from the machinery and equipments, the outdoor noise level, the characteristics of the workers and their hearing deficiencies, and how the workers were initiated with drills, safety warnings and instructions.

What could be the performance measures for warning products in controlled experimentation?

In this case, the investigator needs first, to select objective, accurate, and reliable measurements such as the increase in heart beats above individual baselines, or the duration needed before the heart beat return to its resting level, or the reaction time for any changes in the dilation of the pupil of the eyes, or other physiological characteristics, which are not highly correlated among themselves, if more than one dependent variables are used, and should be used.

Secondly, the purpose in controlled experimentation is to select the characteristic of sound or noise that best affect the outcome of the study in order to design a performing warning alarm with the purpose of eventually reaching design guidelines for products/systems; and third, that the conditions and location of the experiment have to allow the repeatability of the experiment.

More than one session is necessary to sort out the difficulties and distortions in the untrained experimental minds. However, we agreed that in testing the warning alarms we need to control important factors that might affect the results such that we need to test all the selected warning alarms in each one of the facilities, in every location that they might be installed, during the day and night shifts, in the morning, after lunch, and at the end of a shift if necessary and funding is available.