Comments (1)

|

Comments This Article

Human Error in the Management of Building Projects

Andrew Atkinson, School of Construction, South Bank University, London

Introduction

Studies of human error normally are concerned with endeavours with an important implication for safety. Thus, most literature concentrates on accidents and physical failures in such industries as transportation, chemicals, nuclear power generation and, to a lesser extent, civil and structural engineering. In recent years the study of human factors in these industries has been given much impetus by such disasters as Three Mile Island, Chernobyl and the Kansas Hyatt Regency collapse (Reason, 1990; Petroski, 1985).

It is also recognized by some (e.g. Rollings and Rollings, 1991) that human errors are involved in lesser failures, leading to defects, reduced utility, premature ageing and increased maintenance costs for many components or complete structures in all industries. It is with this area of smaller (but more endemic) consequences related to the quality of construction that this paper is concerned.

When human error is considered, frequently it is noted that this predominates as the cause of failure. Percentage figures for the proportion of failures attributed to human error range from 75% (Stewart, 1993) to 90% (Matousek, 1982). This predominance is reported from such diverse industries as process engineering (Hurst et al., 1991), structural engineering (Melchers, 1989), and transportation (Lourens, 1990).

The importance of human error is evidenced also by a notable contention of much of the literature on construction defects avoidance (see e.g. Building Research Establishment, 1991; Department of the Environment, 1989) that defects and their modes of correction, together with the basic sciences involved, are repeated continually and often involve traditional technology as well as newer designs. This supports the view that errors of application rather than technical factors are to blame for most defects.

The lack of significance attributed to technological factors in failures was noted as early as 1976 in the Bragg Report (Health and Safety Executive, 1976). In a consideration of false work failures the report notes: `in hardly any case did we find that failure was the result of a problem beyond the scope of current technology’. If this view is correct, then it is surprising that most literature on the cause and correction of defects and failures continues to concentrate on emphasizing technological solutions. It appears that greater attention should be given to the reasons why known technology is not being applied: in other words, attention to the underlying human errors and the reasons why these errors occur or remain uncorrected. Express consideration of human factors reasons for failures forces attention away from technology and towards psychological, social, or managerial factors which influence the performance and selection of the individual.

Definition of error

It could be argued that all errors are human in origin, but the distinction made in the Bragg Report between technological and human factors implies a definition of human error that excludes so called `state of the art’ failures (Kaminetzky, 1991), where it could not be known that a technology would fail in a particular set of circumstances. Likewise, the effects of `the normal process of deterioration’ (Department of the Environment, 1989) are excluded from that Agency’s definition of failure. An error, therefore, involves an element of individual culpability as, for example, defined by Stewart (1993) as `a human action that exceeds some limit of acceptability’, and by Bea (1994) (in a wide ranging review of the literature concerning error in several industries) as `a departure from acceptable or desirable practice on the part of an individual that can result in unacceptable or undesirable results’.

It is a matter of contention whether individuals can justifiably be blamed for all errors, as making occasional mistakes is seen by many as inevitable (Kletz, 1985; Reason, 1990), but for this paper, the definition supplied by Bea is adopted. The purpose of this paper, therefore, is to investigate the causes of `departures from acceptable or desirable practice’ in relation to construction projects.

Approach of the research

The research reported in this paper is exploratory in nature. It commenced as a review of the literature concerning the causes of human error. Analysis of causes allowed a preliminary grouping into categories, or classes, of cause. These classes, termed `factors’ in the report, in turn, were re-grouped to form a three level hierarchical model. The method of analysis gives the appearance of a relatively mechanistic approach, but this was considered necessary as a preliminary tool and was used to achieve comprehensive coverage rather than sophistication.

The model was tested against a sample of construction industry practitioners with the primary intentions of verifying the grouping (i.e. that the factors were robust) and ranking their relative importance. At the time of the research, some complexities now evident in the study of human error were not obvious. In particular, the inter-relatedness of many causes giving rise to their systemic characteristics has not been covered in detail.

The causes of human error

Individual failings

A consideration of the causes of human error should commence at the level of the individual, all errors having as their basis an individual failing. This invokes the psychology of error, and current thinking in this area is summarized by Reason (1990). Reason draws on Rasmussen (1983) in dividing errors into three types: skill based slips and lapses, rule based errors and knowledge based errors. Skill based slips and lapses stem from the mechanism of the mind for dealing with routine activities and involve such errors as forgetting intentions (perhaps because a delay has intervened between the intent and the execution of an action). Rule based errors stem from another mechanism for dealing with routine performance. For often repeated actions, there is a tendency for the individual to apply rules that have worked in the past. Errors might occur in such situations as where too much information is being received and a wrong rule is applied in panic. Knowledge based errors are those resulting from faults in the deliberate thinking process required for problem solving. This thinking process is very resource intensive and is restricted by such factors as the limitation on the capacity of the mind as a processor.

Two factors in the consideration of errors from the perspective of psychology are first that human errors are, more or less, inevitable, and second that the perspective largely covers only genuine errors, i.e. it excludes errors caused by ignorance, fraud or negligence.

Concerning the first factor, Reason (1990) notes that `Errors are the inevitable and usually acceptable price human beings have to pay for their remarkable ability to cope with very dif® cult informational tasks quickly and . . . effectively’. Concerning the second factor (errors caused by ignorance, fraud or negligence), Reason (1990) does give some attention to the occurrence of `violations’ , that is, `deliberate deviations from those practices deemed necessary to maintain the safe operations of a potentially hazardous system’ .These are caused by the natural human tendency to take the path of least effort, and are aided by a relatively indifferent environment (e.g. an environment that rarely punishes violations or rewards observances). The definition of violations still appears to be very restrictive, and excludes deliberate fraud and negligence.

Technologists writing on the subject of error take a more judgmental view and cite such lapses as `lack of foresight’ (McKaig, 1962) and `laziness, greed, fraud and pride’ (Rollings and Rollings, 1991).

To avoid errors, technologists tend to advocate measures related to the individual, e.g. improving the training, education and selection of personnel (Sriskandan, 1986). Again, however, these suggestions do not really address errors caused by fraud and negligence. Further, Reason (1990) points out those errors, being an intrinsic part of mental functioning, cannot be eliminated by training. He considers controlling the consequences of error to be more productive.

Managerial failings

Solely to emphasize the role of the individual perpetrator in errors, therefore, is problematical. If errors are inevitable, training and education is of limited effect and fraud or negligence is present, no amount of emphasizing the use of correct technology will prevent errors occurring. This fact is fairly widely recognized in the literature on error in the high risk process and transportation industries (Kletz, 1985; Advisory Committee on the Safety of Nuclear Installations (ACSNI), 1993). It is a role of management to ensure that work is carried out properly, to make sure that resources are available at the time they are required and to allocate responsibilities accurately between operatives.

Hierarchy of errors

Closely allied to the limitations of an individualistic view of errors is the realization that disasters, lesser failures and defects often stem from several causes. Turner (1978) suggests that `It is only very rarely that the emergence of a disaster can be fully attributed to the blunders, errors or misunderstandings of a single individual’. This view is repeated by Jones and Nathan (1990) in reporting on a supermarket collapse in Canada, and by Whittington et al. (1992) in considering construction accidents. The latter, recorded that, for the accidents they studied, there were between 3 and 15 causes and an average of 7 per accident.

Hadipriono (1985) divides errors causing failures into `enabling’ and `triggering’ causes and Petroski (1994) introduces the concept of latent failures, which may be triggered by “yet-inexperienced conditions”. Reason (1990) also makes the distinction between latent and active errors and adopts the term “resident pathogen metaphor” to explain latent defects in a system waiting for a combination of circumstances to occur. Blockley (1992) proposes a `balloon model’ of latent precursors to an event corresponding to air being blown into a balloon. Subsequently, only a small trigger event (the active error), such as a pin or lighted match, is needed to release the energy pent up in the system.

The concepts of latent and active errors are developed further by authors writing both from a reliability engineering (Embrey, 1992) and a construction management perspective (Whittington et al., 1992; Eldukair and Ayyub, 1991). These authors adopt hierarchical models of the error process in complex socio-technical systems. Of particular relevance to construction projects is the division of error causes by Eldukair and Ayyub (1991) into primary and secondary classes and by Whittington et al. (1992) into immediate causes, site management issues and headquarters issues. A common feature of most of these models is the division of causes into three categories.

1. Causes related to the individual (i.e. the active error which precipitates the failure, accident or defect).
2.  Causes related to managerial ineptitude such as lack of supervision and control.
3. Causes related to wider factors such as the economic climate, pressures of time and political constraints.

Although Eldukair and Ayyub (1991) make the distinction between primary and secondary causes in construction projects and (based on the analysis of published failure reports) make some attempt at apportioning cause of failure to personalities involved, they stop short of conducting a full analysis of managerial and wider causes. However, a detailed analysis of the literature encompassing defects, failures and accidents (and their corollary, quality and safety) indicates that some hierarchical grouping is possible.

A preliminary model of the error process in construction projects

The literature covering defects, failures, accidents, safety and quality assurance related to project based industries suggests that it is possible to construct a model of the error process specifically related to construction projects. This could be based broadly on a three-level hierarchy consisting of primary causes, secondary or managerial causes and tertiary or global causes. Primary causes relate to factors relevant to the individual. Managerial causes consist of a range of factors, under the control of the managers of project based organizations. Global causes relate to factors which loosely could be classed as outside the control of the organizational unit.

Primary causes

As already noted, these causes stem from internal psychological processes or result from ignorance, negligence and fraud. To control primary errors at source requires proper knowledge acquisition in the form of education, training and experience, adequate conditions for carrying out work and/or the selection of knowledgeable personnel. Also suggested (by e.g. Jones and Nathan, 1990) is the use of self-inspection of tasks by the individual. These measures may control genuine errors to an extent, but will still be ineffective against deliberate fraud or violations.

Managerial causes

Failings in checking, supervision and control-By far the most widely quoted managerial measure for containing primary errors is the use of checking in one form or another. The effectiveness of checking in relation to structural engineering calculations has been demonstrated by Stewart and Melchers (1989). If not the actual term checking, then the looser terms of supervision, inspection and control are widely cited (see e.g. Kaminetzky, 1991).

Despite its widespread advocacy, the use of checking and inspection suffers from three limitations. First, checking is intermittent and cannot be expected to detect all primary errors. This is pointed out by several authors including McKaig (1962) and Kaminetzky (1991). Second, checkers frequently make the same errors as the original perpetrator, thus rendering the process ineffective (Jones and Nathan, 1990; Petroski, 1994). Third, checking assumes that errors `percolate’ upwards from the work-face. In fragmented project based industries such as construction, the errors are as likely to come from the checkers (in the form of `design’ or `managerial’ errors) as from the construction workers themselves. These limitations indicate both that checking alone will not remove all errors and that error prevention exhibits systems characteristics, depending on the interaction of primary, secondary and tertiary factors.

Confusions of responsibility - Daoud and Hamdani (1988) consider that the lack of a clear understanding of the rights and responsibilities of each party in a project is a major source of problems in construction. This theme is taken up by Jones and Nathan (1990) and Eldukair and Ayyub (1991). The latter report that, in their review of press reported failures, deficiencies in work responsibilities contribute to about 30% of cases and are the dominant `management’ cause. To remove confusions of responsibilities, written procedures are suggested. Authors from a forensic engineering background (Feld, 1968; Kaminetzky, 1991) also advocate the removal of split responsibilities by, for example, concentrating responsibility for design in the hands of one architect or engineer (as opposed to, say, having one designer for the `scope’ elements of a project and another for the `detail’ elements).

Errors caused by changes - The dynamics of change during the design and construction period have been cited in several major construction failures, including the collapse of a walkway in the Hyatt Regency Hotel, Kansas, USA in 1981 in which 150 people were killed (Petroski, 1985). Changes are also implicated in less spectacular failures and in the generation of defects and accidents (Brown and Xiaochen Yin, 1988; Daoud and Hamdani, 1988).

The project management concept of change control (Morris, 1994) is invoked in controlling change including the use of proper planning, written change control procedures and effective communication of changed requirements.

Errors caused by concurrency - The term `concurrency’ is not used widely in the error literature reviewed but, taken from general project management (Morris, 1994), it means the simultaneous execution of several phases of a project. The concept of concurrency, although not often mentioned expressly, is cited frequently by implication in the literature as a cause of errors. Chevin (1993) quotes Graham Matthews of the British Airports Authority in deprecating the `evolving brief’, where construction is made to follow changes of mind induced at a late stage by the client. Chadwick (1986) does mention concurrency expressly in noting that `Concurrent design and construction frequently cause expensive and embarrassing rework with adverse impact on quality, particularly if these two functions are being carried out by separate organizations’.

Failures in communications - Poor communications, both formal and informal, are mentioned widely as a cause of construction failures. In relation to failures in formal communications, poor drawings and specifications are often cited (e.g. by Brown and Xiaochen Yin, 1988 and Kaminetzky, 1991).

Regarding informal communications, Kletz (1985) notes problems in the construction of the West Gate Bridge, Victoria, Australia, which led to its collapse in 1976: `No one told (the construction team) that (box girder) components must not be forced together, if they do not ® t they must be modified. The consulting engineers made no attempt to ensure the contractors understood the design philosophy and that traditional methods of construction could not be used. Nor did they check the construction to see that it was carried out with sufficient care’. The last sentence illustrates the close link between some failures of communication with failures in checking. In this case, checking appears not to be to detect `culpable’ errors, but to ensure that the construction team was briefed adequately.

Turner (1978) suggests several specific reasons for communications errors, including non-receipt, distortion, ambiguity and overwhelming volume of information, and a lack of overview concerning communications.

Several suggestions are made in order to improve communications, including the better use of written specifications and plans (Dea and Gans, 1986). Both Reason (1990) and Norman (1988), from the perspective of psychology, emphasize the importance of making procedures visible, so that the path of communications is clear.

Global causes

Financial pressure as a cause of error - Financial pressure as a cause of error is mentioned by several writers including Chadwick (1986), Brown and Xiaochen Yin (1988) and Petroski (1985). The pressures include overemphasis on first cost at the expense of lifecycle costs, inadequate funding of pre-contract investigations, over-keen tendering by contractors and over tight fees for professional services including post contract inspections. Whittington et al. (1992), from an industrial safety perspective, criticize financial pressures, leading to a preference for short term contracting and reductions in safety standards in the UK building industry.

Although the view that cost pressures lead to errors is expressed widely in the literature, a contrary view is also proposed by Blockley (1992) and ACSNI (1993). ACSNI points to research reviewed in their publication which indicates that there may be a link between a low error rate and increased economic efficiency, in other words an intervening factor (possibly related to management style) may both improve performance and reduce costs.

Time pressures as a cause of errors - Pressure of time as an underlying cause of errors is noted by Petroski (1985), Brown and Xiaochen Yin (1988) and Rollings and Rollings (1991). Two reports of a failure of a beam in an airport building drawn from Engineering News Record (Korman, 1991a, b) note factors of overloading of work and pressure to produce. The Commission of Inquiry examining the Summerland ® re of 1974 (resulting in the death of over 50 people) and noted in Turner (1978) implicated time and the pressure of work to meet the start of the tourist season as a cause of that disaster.

Contrary comments relating to cost pressures apply also to time: it is possible that lack of time may not in itself be a cause of error, but that good time performance may also be associated with low error rates.

Cultural and social pressures as a cause of errors - Considerable comment exists in the literature on the role of organization culture and wider social issues in the generation of errors. However, although the concept of culture is widely cited, its exact definition often is treated vaguely or not addressed at all. For example, Reason (1990) lists general failings of incompetent management, selective blindness, conflicting goals, reversed logic and fallible management structures, but without identifying specific problems. General comments on cultural and societal factors are also made by Brown and Xiaochen Yin (1988) and Eldukair and Ayyub (1991).

Turner (1978) closely examines the social context of several major disasters and identifies the conditions within various organizations which allowed the disasters to develop. These conditions include rigidities in perception within the organization, remoteness of top management and organizational exclusivity (i.e. the disregard of and lack of interest in views and issues arising from outside the organization).

Summarizing the consensus view from the above authors it appears that the principal indicators of positive organizational culture are

• Visible senior managers,
• Participative and approachable senior staff,
• Involved senior staff,
• Ownership of actions directed towards quality or safety by the workforce, and
• Consensus based production orientation.

Tests of the model

A model of the error process in project based industries therefore can be summarized as follows.

• Primary factors 
• Knowledge acquisition
• Training
• Education
• Experience
• Selection
• Self-inspection 
• Managerial factors
• Checking, inspection and control
• Division of responsibilities
• Change control
• Control of concurrency
• Communications 
• Global factors
• Cost
• Time
• Organizational factors (culture)
• Societal pressures

In order to test this model, a questionnaire survey was conducted of construction industry practitioners. The specific objectives of this survey were to:

• Gauge the relative importance of the factors in the model,
• Extend the range of factors identified or, alternatively, to confirm that the model was adequately comprehensive,
• Test the general applicability of the model within the construction industry and examine differences between broad sectors of the industry, and
• Confirm the `three-layer’ structure of the model.

Target population requirements were that respondents were drawn from (1) the design and briefing sectors of the construction industry, (2) the construction sector of the industry and (3) a relatively neutral sector. The reason for the three divisions was to detect any sharply polarized views. If views between designers, constructors and the neutral respondents were reasonably correlated, it might be concluded that responses had not been affected unduly by polarized attitudes.

To fulfil the role of neutral respondent it was decided to target consultant or client based quantity surveyors. It was recognized that consultant quantity surveyors are normally paid by the client and thus are not strictly neutral, but characteristically they occupy a more detached position between the design and construction elements of projects than either the constructor or designer. In particular, they are not involved directly in the design or construction process, and they are required by most standard forms of contract to take a neutral stance in matters of post-contract evaluation.

The sample consisted of construction industry participants recruited by final year students of quantity surveying, building surveying and construction management following day-release honours degree courses at the South Bank University. This was not a strictly random sample, and responses were drawn largely from the south east of the United Kingdom. Additionally, some design professions (architects/engineers) were not approached directly, and respondents were drawn from relatively established organizations.

420 questionnaires were distributed and 107 were returned, representing a 25.5% return. The questionnaire consisted of a 28 item three page `closed question’ style instrument, including 14 questions asking for both a rating (out of 5) and a ranking of methods of avoiding errors leading to defects. The respondents were asked to rate every method, but only to rank the top ® ve methods. The reasons for asking for both a rating and a ranking were to check for an understanding of the questionnaire and to check for random completion. The 14 questions covered the factors identified in the model, and an opportunity was also given in question 15 for respondents to supply further factors. Questions were ordered to try and avoid primacy and recency effects in completing the questionnaire.

Analysis

Table 1 Role of respondent in detail

ROLE	NUMBER	PERCENTAGE
ARCHITECT	4	3.7
BUILDING SURVEYOR	18	16.8
ENGINEER	6	5.6
OTHER DESIGNER	27	25.2
QUANTITY SURVEYOR	25	23.4
MAIN CONTRACTOR	21	19.6
SUB CONTRACTOR	6	5.7
TOTALS	107	100

Role

The roles of respondents as classified by their employing organizations were as shown in Table 1. Using the three ways grouping identified above, the sample of 107 consisted of 55 designers (51.4%), 27 constructors (25.2%), and 25 quantity surveyors (23.4%). Respondents’ experience amounted to an average of 17.2 years in the construction industry, including an average of 5.8 years in their current roles.

Supervisory role

Respondents also were asked whether or not they supervised other staff within their organization. This was in order to try and detect whether responses were distorted by a `blame’ or `attribution’ effect. These effects are discussed in more detail later. 63.6% of the sample supervised the work of others, 32.7% did not and 3.7% did not respond to this question.

Additional factors in the model

Respondents were given the opportunity in the supply question (15) to indicate other factors that they considered important in avoiding errors. Of the 107 returns, 51 made comment under this question and only 2 made points that could not be incorporated readily into the factors identified above. There was a strong indication in the responses that factors loosely termed `managerial’ (covering aspects such as leadership and motivation) and relating to systems and culture were considered important. By contrast, the specific factors of communications, supervision, time and economic influences were less widely mentioned than evidenced in the ratings for questions 1± 14.

The applicability of the model in the construction industry

The responses were analysed to determine the extent of agreement between the main groups of respondent, designers, constructors and quantity surveyors. A further analysis was undertaken between those supervising and those not supervising the work of others. Here the possibility of a blame or attribution effect was being examined. Blame effects might distort the responses by leading to an overemphasis on individually based factors at the expense of managerial and external factors. Whittington et al. (1992) report the effect as `Assigning responsibility to individuals absolves the supervisor and company from direct blame and typically leads to recommendations for behaviour change often through disciplinary approaches’ .

Attribution theory is drawn from psychology and, similarly, suggests that the individual might blame external circumstances for his/her own inadequacies, but blame the failings of others on personal inadequacies. Heider (1958) defines attribution theory in terms of `an overall tendency has been found to

attribute what happens to other people to their personal responsibility, but to place much more emphasis on the role of external circumstances, when explaining what has happened to oneself ’ . A blame or attribution effect might distort responses and introduce inaccuracies in the model. If widespread, it might give an insight into reasons for an over-emphasis evident in some literature on primary and technological factors at the expense of managerial factors.

All data were treated as ordinal in the analysis, and tests between groups of the relative importance of factors using Kendall’ s tau and Spearman rank correlations revealed significant (p < 0.05) agreement. This was the case between both groups divided by role and groups divided by supervisory position.

Differences between groups for particular factors

For some separate factors it was apparent that differences in view were held. To test this a Mann-Whitney

U± Wilcoxon rank sum W test was conducted for each factor within the two broad divisions, role and supervisory position.

Discussion of the differences

Differences between roles - Differences of factor ratings between the roles of respondents was considered a possibility, but not predicted in the survey. Accordingly two-tailed tests of significance were used. These indicate that most illustrated differences between designer and quantity surveyor are not significant at the p <0.05 and one (communications) is not significant at the p < 0.10 level. Thus, these differences may be due to chance. Even where close to achieving significance (for example for the factor responsibilities), it is dif® cult to draw plausible conclusions from the differences. It is possible that quantity surveyors take a greater interest in and place more emphasis on the bureaucratic functions of ensuring good communications, careful division of responsibilities and the careful selection of sub-ordinates. They might also have a greater awareness of the importance of time in performance. However, to support these suppositions a corresponding emphasis might be expected by designers on other factors (for example, training, education, culture). This was not apparent.

The differences between designer and constructor are much more marked, with two-tailed significance (p < 0.05) being achieved on four factors, changes, communications, concurrency and economics. On the global factor economics constructors rate the influence of the economic climate significantly higher than both designers and quantity surveyors. The constructors’ positions as commercial concerns (as opposed to designers/quantity surveyors more consultative roles) may account for this. However, given current pressures on fee income and public client market testing of services, a significant difference in the perception of economic influences is, perhaps, surprising.

In common with the quantity surveyors, constructors rate some managerial factors more highly than do designers, but, again, there is no corresponding emphasis by designers on other factors. Constructors, as represented by contracting and sub-contracting firms, are, however, much more involved than designers in the integration and control of diverse resources, and it is this involvement that may account for their greater emphasis on managerial factors.

Overall, there was less difference of views between quantity surveyors and both constructors and designers, than there were between designers and constructors. This matches with the assumption in the design of the survey that quantity surveyors sit in a `quasi-neutral’ position between the construction and design parts of the project processes.

Differences between supervisors and non-supervisors

The significant difference in the direction opposite to that predicted on the factor `supervising the work of junior staff’ tends to discount the blame or attribution effect while significant differences were not detected elsewhere, and thus a contrary perceived readiness to accept responsibility on the part of supervisors cannot be claimed from the results.

The broad structure of the model

To examine the possibility that the factors identified in the model cluster into the three level hierarchy, or on other underlying factors not as yet identified, the statistical technique of factor analysis was used. To facilitate this possibility, the questionnaire intentionally did not include any indication of overall grouping of the questions. Should any emerging factors correspond with the model, this would provide considerable support for its robustness.

Factor analysis is a technique used in the study of psychology to try and detect underlying latent variables as displayed by scores on other variables. Using SPSS for Windows ± - factor analysis procedures (professional statistics option) (SPSS, 1995; Kinnear and Gray 1994), an initial correlation matrix was constructed using all 14 questions from the questionnaire. The factor `self-inspection of tasks’ was excluded from the analysis as it had no significant correlations with any other factor. Using procedures recommended in Kinnear and Gray (1994) the factor time was later omitted from the analysis as it correlated too closely with all other factors. Finally, repeating this procedure, the factor responsibilities were omitted from the analysis, giving a reduced matrix of 11 variables.

Discussion of the factor analysis

In all levels of the factor analysis, the factors of `economics’ , `client expectations’ and `politics’ clustered as a group (3). This is approximately, but not exactly, in line with the model. All three factors were given relatively low ratings in the returns. The clustering of these factors suggests that the model could be modified to reconfigure `global factors’ to contain these three sub-elements together, perhaps also including the factor `time’ , which had a relatively high correlation with several factors, but exhibited characteristics of an overall global driving force.

In the final factor analysis, the factors in latent factor 1 followed fairly closely the group `managerial factors’. However, this analysis excluded `responsibilities’ and `time’. The clustering of the factors suggests an identifiable class of managerial factors, but should, perhaps be modified to include `organizational culture’ and `responsibilities’, but exclude `time’. The justification for including `responsibilities’ as a managerial factor is that, logically, allocating responsibilities is managerial in favour, even though this is not supported directly by this analysis.

`Education and training’ and `selection of individuals’ as factors clustered together throughout the analysis, and support the grouping in the model of `primary influences’. The factor `self-inspection of tasks’ was omitted in the analysis, but it was not rated highly in the returns. It is suggested that `self-inspection’ is perhaps really part of the knowledge of the individual and can be removed as a separate factor.

`Communications’ alone among the factors clusters with more than one group. This might suggest that respondents regarded communications as being both a function of management and an individual skill or knowledge. However, given the purely statistical basis of the analyses, this duplication should not be overemphasized.

Overall, the factor analysis can be taken to support the model. The structure of the model was determined primarily on the basis of the literature, and does not rely on the statistical analyses. However, it is noteworthy that such a consistent loading on pre-identified latent factors was found.

Conclusions

The object of the survey was, primarily, to gain support for the range of factors previously indicated in the literature on human error and to obtain a view of the relative importance of these factors. The range of factors does appear to be reasonably robust, but it should be noted that the survey participants were faced with a predetermined menu. Question 15 (the open question) sought to overcome this criticism by inviting inclusion of other factors. The lack of wholly original supplements to those offered gives some measure of confidence in the model. However, even allowing for the inclusion of question 15, it is clear that participants were `led’ somewhat. This illustrates a general problem with postal style, relatively anonymous questionnaires: a wholly undirected, open questionnaire would probably be too dif® cult to analyse, whereas a closed instrument can be criticized for being too pre-determined. Nevertheless, the results, when combined with published literature suggest that the model has some validity.

With respect to the relative importance of factors, it is noteworthy that managerial factors of `communications’ , `concurrency’ , `change control’ and `checking’ all rank in the first six. However, `time pressures’ (a global factor) and `education and training’ (a primary factor) are also considered highly influential. This suggests that error prevention is a team effort and has socio-technical, or systemic characteristics as identified by (amongst others) Embrey (1992) and Blockley (1992). Systems characteristics, in particular the causal links between factors, specifically were not tested in the research, and these findings point to further work and refinements to the model.

The views of respondents are fairly uniform when a division on three broad occupational groups is considered. Constructors as a group tend to emphasize managerial factors more than designers, and quantity surveyors sit somewhere between these two groups. It is suggested that the relative emphasis placed by constructors on management stems from their greater involvement in planning and organization of resources: in other words, they are more involved in detailed project management. In view of the fact that designers often take on duties of overall project direction, their relative myopia could be contributing to some errors. Supervisors rated the function of `supervising the work of juniors’ much more highly than non-supervisors, but this was opposite to an expected blame effect.

The overall structure of the model is tentatively supported by the survey responses in that global, managerial and primary factors do appear to fall into groups. Factors falling in both managerial and primary groups were strongly supported, but global factors of political influences, client expectations and economic pressures were less so. This suggests that respondents considered performance in relation to errors to be in their own hands and not particularly susceptible to external influences.

Further work

The results of the survey give support to the structure and overall content of the proposed model, but further work is indicated. Research investigating exactly how errors develop or, alternatively, how individuals and organizations handle problems in avoiding errors would be valuable in guiding project based industries towards such objectives as better standards of quality and lower accident rates. The systemic nature of error commission and avoidance was noted, but not covered in detail. Further work would be necessary to investigate how factors interact. This work would probably involve direct observation of participants and projects and measurement of project performance. Direct observation and measurement presents some interesting problems. These include inaccessibility of data, particularly related to culpable errors and failures, difficulties in measuring errors, the large number of factors to consider, and

the intangible nature of many of the factors.

One key benefit of the current study is that it indicates priority areas for further work, perhaps starting with a detailed examination of the influences of the most highly rated factors ± communications, time pressures, the control of concurrency and project changes, education and training, and the interactions between them.

References

• Advisory Committee on the Safety of Nuclear Installations (ACSNI) ± Study Group on Human Factors (1993) Third Report: Organising for Safety, HMSO, London.
• Bea, R.G. (1994) The Role of Human Error in Design, Construction and Reliability of Marine Structures, U.S. Ship Structures Committee. 
• Blockley, D.I. (1992) Engineering Safety. McGraw-Hill, New York. 
• Brown, C.B. and Xiaochen Yin (1988) Errors in structural engineering, ASCE Journal of Structural Engineering, 114(11), 2575± 93. 
• Building Research Establishment (1991) Housing Defects Reference Manual: The Building Research Establishment Defects Action Sheets. E & FN Spon, London. 
• Chadwick, W.L. (1986) Impact of design, construction and cost on project quality, ASCE Journal of Professional Issues in Engineering, 112(2), 69± 79. 
• Chevin, D. (1993) Management of design, Building Magazine, 5/11/93, 36± 41.
• Daoud, Osama E.K. and Hamdani, S.K. (1988) Human mistakes in construction ± case study in the Gulf region. Forensic Engineering, 1(3), 175± 84. 
• Dea, S.J. and Gans, R.L. (1986) Design standardisation program of WSCC. Journal of Management in Engineering, 2, 111± 23. 
• Department of the Environment (UK), Property Services Agency, Directorate of Building and Development (1989) Defects in Building, HMSO, London. 
• Ebbinghaus, H. (1885) Memory: A Contribution to Experimental Psychology, Rev. Edn. 1964. Dover, New York. Quoted in Hayes, N. (1994) Foundations of Psychology. Routledge, London. 
• Eldukair, Z.A. and Ayyub, B.M. (1991) Analysis of recent U.S. structural and construction failures. Journal of Performance of Constructed Facilities, 5(1), 57± 73.
• Embrey, D.E. (1992) Incorporating management and organisational factors into probabilistic safety assessment. Reliability Engineering and System Safety, 38, 199± 208. Feld, J. (1968) Construction Failures. Wiley, New York. 
• Hadipriono, F.C. (1985) Analysis of events in recent structural failures. ASCE Journal of Structural Engineering, 111(7), 1468± 81.
• Heider, F. (1958) The Psychology of Interpersonal Relations. Wiley, New York. Quoted in Lourens (1990).
• Health and Safety Executive (1976) Final Report of the Advisory Committee on False work (Bragg Report), HMSO, London.
• Hurst, N.W., Bellamy, L.J., Geyer, T.A.W. and Astley, J.A. (1991) A classification scheme for pipe work failures to include human and socio technical errors and their contribution to pipe work failure frequencies. Journal of Hazardous Materials, 26, 159± 86.
• Jones, P. and Nathan, N.D. (1990) Supermarket roof collapse in Burnaby, British Columbia, Canada. Journal of Performance of Constructed Facilities, 4(3), 142± 60.
• Kaminetzky, D. (1991) Design and Construction Failures ± Lessons from Forensic Investigations. McGraw-Hill, New York.
• Kletz, T.A. (1985) An Engineer’ s View of Human Error. Institute of Chemical Engineers, Rugby, UK.
• Kinnear, P.R. and Gray, C.D. (1994) SPSS for Windows Made Simple, Lawrence Erlbaum Associates, New York.
• Korman, R. (1991a) Defence attacks splintered process. Engineering News Record, 226, 29 April, 8± 9.
• Korman, R. (1991b) Penalties point to design errors. Engineering News Record, 226, 11 February, 8± 9.
• Lourens, P.F. (1990) Theoretical perspectives on error analysis and traf® c behaviour. Ergonomics, 33(10), 1251± 63.
• Matousek, M. (1982) Measures against Errors in the Building Process. Institute of Structural Engineering, Swiss Federal Institute of Technology, Zurich, Switzerland (Translation No. 2067, National Research Council of Canada, 1983).
• McKaig, T.H. (1962) Building Failures ± Case Studies in Construction and Design. McGraw-Hill, New York.
• Melchers, R.E. (1989) Human error in structural design tasks. Journal of Structural Engineering, 115(7) 1795± 807.
• Morris, P.W.G. (1994) The Management of Projects. Thomas Telford, London.
• Norman, D.A. (1988) The Psychology of Everyday Things. Basic Books, New York.
• Petroski, H. (1985) To Engineer is Human. The Role of Failure in Successful Design. Macmillan, London.
• Petroski, H. (1994) Design Paradigms ± Case Histories of Error and Judgement in Engineering, Cambridge University Press. 
• Rasmussen, J. (1983) Skills, rules, knowledge: signals, signs and symbols and other distinctions in human performance models. IEEE Transactions: Systems, Man and Cybernetics. 
• Reason, J. (1990) Human Error, Cambridge University Press. 
• Rollings, R.S. and Rollings, M.P. (1991) Pavement failures: oversights, omissions and wishful thinking. Journal of Performance of Constructed Facilities, 5(4), 271± 86. 
• SPSS (1995) SPSS for MS Windows, Release 6.1 SPSS, Inc. 
• Sriskandan, K. (1986) Quality Assurance under Different Forms of Contract. International Association for Bridge and Structural Engineering, Report No. 50 (Symposium ± Tokyo, Japan 1986), pp. 41± 50. 
• Stewart, M.G. (1993) Structural reliability and error control in reinforced concrete design and construction. Structural Safety 12, 277± 92. 
• Stewart, M.G. and Melchers R.E. (1989) Checking models in structural design, Journal of Structural Engineering, 115(6), 1309± 24.
• Turner, B.A. (1978) Man Made Disasters, Wykeham, London. 
• Whittington, C., Livingston, A. and Lucas, D.A. (1992) Research into Management, Organisational and Human Factors in the Construction Industry. HSE Contract Research Report No. 45/1992. HMSO, London.