• Tidak ada hasil yang ditemukan

Data collection for task analysis

Dalam dokumen Human Factors for Engineers (Halaman 110-113)

Task analysis

5.2 Data collection for task analysis

In any task analysis it is first necessary to collect data that will determine how tasks are done and then to supplement these data with more focused information about aspects of the tasks, or the task environment, that are of particular interest. Inevitably this will require that the analyst will have to use a variety of sources. This means that on occasion it will be necessary to resolve potentially conflicting data, or to collate several data items to clarify ambiguities. Thus, the data collection prob- lems are similar to those that may be met in the analysis of any complex system in which many of the details are either difficult to obtain or are open to different interpretations.

However, there are also some aspects of data collection that are particularly impor- tant for task analysis and so the following brief summary of some of the main issues is presented, based upon the nature of the information source.

5.2.1 Existing documentation

Even in a new system, a vast amount of documentation is often available about tasks and task interfaces, including such things as functional descriptions, system descriptions, logic diagrams, piping and instrumentation diagrams, interface layout diagrams, training programmes and written procedures. In existing systems this could be supplemented by historical information about past problems, such as operator logs and critical incident reports.

However, whilst this information is useful for task analysis purposes, it is important to recognise where there may be potential inaccuracies. In a review of several task analyses undertaken for a wide range of applications(summarised in [1]) a major source of such inaccuracies was identified as being due to either ignoring many features that could influence human performance, or making implicit assumptions that are not documented. This is particularly true for system descrip- tions, because these tend to focus on the functional requirements rather than the human requirements. Thus, features that can have significant implications to human performance may either not be documented, or decisions may be made without properly considering their potential impact. For example, whilst a designer is likely to define the electrical specification for a particular switch, other features that may directly affect the way that the switch is used, such as its accessibility or the feedback that it provides, may not be specified. If an inappropriate switch is then installed, an operator may subsequently make errors because the status of that switch is misidentified, or because it is difficult to reach during an emergency.

In order to guard against such misinterpretation, an analyst must always be care- ful not to extrapolate beyond what is actually specified. Also, where possible, all information should be corroborated from a different source.

Written procedures and training documentation can both be particularly useful for developing task descriptions. However, the analyst must be aware that some steps may be covered very briefly, or omitted altogether. This is particularly likely in training

notes, because these will tend to focus upon aspects that tend to cause difficulties for trainees, or that have safety significance.

Operational feedback records and incident reports can provide valuable informa- tion about features that have caused past difficulties, and which, therefore, may well merit special attention.

5.2.2 Interface inspection

If the control interfaces, or a simulation of them, have already been built, much information about those interfaces can be obtained from making a systematic survey of them. This could be done as part of a walk-through exercise in which particular controls and displays are examined as they become used during a task sequence, but it is often preferable to undertake a systematic inspection of an entire interface, either by working from the top left to the bottom right of each control panel, or by systematically working through a set of VDUdisplay formats.

Such inspections should consider compliance with basic ergonomics guidelines such as the consistency of labelling(terminology, size and relative position), the grouping of items, legibility(of labels and parameter values), coding consistency, the adequacy of feedback, direction of motion stereotypes and control/display compati- bility. For important indications, that may be more remote from the main operating position, it may also be necessary to check the sightlines to ensure that both the short- est and the tallest operator have unhindered visibility of such items without having to move from their normal working positions.

For interfaces that have been operational for some time, it can also be useful to undertake an operator modifications survey to look for areas where the operators have felt it necessary to supplement the interfaces by addingDymo tape labels or marking critical limits on displays.

5.2.3 Subjective assessment

A considerable amount of useful task information can be obtained by individual or group discussions with operational personnel or technical experts. However, there is always a risk that such information could represent a distorted, or even inaccurate, view. Therefore, an analyst must take considerable care to ensure that their inter- pretation of such discussions is accurate and to recognise situations where there are different ways to undertake a task.Problems of misinterpretation can be reduced by providing an opportunity for respondents to comment on an analyst’s interpretation, either during a discussion session, or by undertaking a validation of the analyst’s inter- pretation at a later date. The importance of such validation cannot be overstressed because it is surprisingly easy for an analyst either to miss out an important aspect of a task, or to derive an ambiguous or erroneous view of the task by misinterpret- ing an interviewee’s accounts or opinions. It must always be remembered that what may appear straightforward to a highly experienced operator may in reality involve some highly complex cognitive behaviour and decision-making. Thus, whilst there is a tendency to obtain task data from the most experienced operators, in some cases,

a less experienced operator may be a better source of information about aspects of the task that eventually become highly skilled and undertaken with little apparent conscious effort.

The best way to identify alternative approaches to undertaking a task is to have discussions with more than one operator and then to compare their approaches. How- ever, care must be taken to select the most appropriate respondents. The author recalls a situation where the most experienced operator described how an additional pump would be brought into operation and this approach was later corroborated by another person on the same shift. However, further investigation with a relative newcomer revealed that there was a better way to do this task, which involved less thermal shock to the system. The experienced operator was not aware of this problem, and so for several years he had continued to add pumps in this way and to train others to adopt the same, suboptimal, approach.

In order to get the most out of group discussions it is advisable to structure such discussions upon a particular procedure, or to undertake them as a debrief session after a particular task has been completed. It is also helpful if those attending group discussions can be pre-warned of the topics that will be covered. In any discussion group there must be an appropriate mix of individuals with the necessary expertise and, if there is to be a series of discussion sessions, it will be helpful if the same individuals can be used. To encourage discussion whilst avoiding certain views pre- dominating, only one expert should generally be permitted from each area of expertise.

There should also not be too great a difference between the managerial levels of the members of a discussion group, otherwise, there is a strong risk that the views of senior management will dominate.

5.2.4 Task observations

Task performance can be observed directly during walk-throughs, simulator trials, or whilst tasks are being carried out on the real plant. Walk-throughs can be under- taken in a similar manner to the actual tasks, apart from actually intervening with the plant, or else they can be interrupted by the analyst so that detailed questions can be asked. All of these data collection methods should provide the analyst with sufficient information to develop a comprehensive set of task descriptions and to identify the controls and displays that are used. They can also provide a way to identify potential problems, such as difficulties locating particular information.During any task obser- vations the analyst must be aware of ergonomics guidelines and note any situations where these are not being met. Sometimes these will be obvious, because an oper- ator will experience some difficulties, but in other cases this may be more difficult because an operator has learned to overcome such deficiencies. Nevertheless, even when operators experience no apparent difficulty with a feature that does not comply with ergonomics guidance, this should still be noted, because these features may cause difficulties to less experienced personnel, or to anyone under more stressful conditions.

The debriefs or the walk-through questions also provide a opportunity to explore the underlying decision-making and reasoning, and also to identify alternative sources

of information, or corroborating sources of information, such as upstream flow meters to confirm a valve position.

Dalam dokumen Human Factors for Engineers (Halaman 110-113)