Ergonomics

Ergonomics

Some case studies

There are many psychological studies which give insight into worker performance, some examples of which are given in the following paragraphs.

Postcode

The (UK) Post Office employs an ergonomics department and has done for many years. As is known, the postcode (e.g. TR27 5JZ) gives a unique reference for each post delivery round in the whole country; the actual combination of letters and numbers is easier to remember than an equivalent string of only digits or letters.

Shiftwork

A survey in Japan of 17000 shift workers in 1500 firms results in recommended 40 hour weeks, overtime limit of 150 hours per year, two-hour sleeping periods during night shifts, not more than eight night shifts per month, etc.

Similar studies in Europe and elsewhere have highlighted that morning-active people differ considerably from evening-active people in how much sleep they got during shiftwork. These two types are easy to identify psychologically and ought to be borne in mind when recruiting for shiftwork.

Production line inspection

Most products require visual inspection to ensure adequate quality, e.g. the glass industry. The signal detection theory provides the statistical background whereby the four possible conditions can be worked out. These choices can also be shown in graph form. (Graph contained in PDF.)

The shaded area of overlap represents the area of uncertainty. By knowing the data for the four possible conditions a single figure can be worked out for the inspector's criterion by using signal detection theory; also the probability of good or reject items are obtained.

These two figures enable management to assess much more accurately the costs and benefits of product quality, bearing in mind cost of inspection and scrapping or salvage of defective items. There is little doubt that user-friendly computer software to perform the statistical calculations of signal detection theory is or will soon be available for management services; thus taking the sting out of what appears superficially to be complicated statistics.

The improvement of inspection has been aided by using a complicated apparatus, which fits onto the inspector's head; this contains a small television camera which records the movement of a light reflected from the eye, superimposing it on what the inspector is looking at. In this way a video film is obtained which shows exactly what the inspector is looking at.

From this, the search pattern, or strategy, used by the inspector in searching for product defects is obtained. For example, a study of 20 inspectors of production of small roller bearings found that the better inspectors looked directly at 85 per cent of the visible bearing surface whilst the worse inspectors looked directly at only 30 per cent, placing a greater reliance on the less-accurate peripheral vision.

It is reported that better training of inspectors and running the belt towards the inspectors instead of away reduced defects by 25 per cent and inspection time by 50 per cent.

Reaction times

With increasing complexity and cost of machines, the need for an operator to react quickly to correct a malfunction becomes increasingly more important. Donders, over one hundred years ago, correctly identified three types of reaction requirement:

  • Type A
    Has an event/signal occurred? (yes/no).
  • Type B
    Which event/signal has occurred and therefore which response is appropriate?
  • Type C
    Which event/signal has occurred and therefore is a response necessary?

From these, the discrimination time = type C response time minus type A response time and choice time = type B response time minus type A response time.

These are used today in information processing. Like computers, information for humans is measured in bits; obviously a single choice is one bit, i.e. either 'on' or 'off'.

The statistics concerning the likelihood of an event/signal are a little involved but relatively simple, e.g. if there are two events/signals which occur with probabilities of 70 per cent and 30 per cent respectively, the average information is 0.88 bits (the average of 0.51 and 1.71 bits worked out from log21 divided by each probability). It is thought that our brain cannot process more than about 12 bits a second. Reaction-time tasks involve a decision based on successive samples of sensory information. Each sample contains some extraneous information ('noise') as well as relevant information.

Next | Indicators (for the application of ergonomic study)