Improving Safety[1]

of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
IMPROVING SAFETY BEHAVIOUR 5 Improving safety behaviour using goal setting and feedback T.W. Marsh, I.T. Robertson, A.R. Duff, R.A. Phillips, M.D. Cooper and A. Weyman A description of the development and effects of behaviourally-based management techniques in improving construction site safety Introduction As in most countries worldwide, the UK construction industry has an unenviable safety record, with construction workers approximately three times more likely to suffer serious injury than
Document Share
Document Tags
Document Transcript
  Introduction As in most countries worldwide, the UK constructionindustry has an unenviable safety record, withconstruction workers approximately three times morelikely to suffer serious injury than other workers. On pastevidence, it is likely that in the UK about 150 people willdie by accident each year and a further 2,500 to 3,000 willbe seriously injured. In addition, 30,000-40,000 will sufferlost-time injuries of more than three days (Davies andTomasin, 1990). A review of previous research revealsthat attempts to improve site safety have had limitedsuccess. These attempts have included “blitz” inspectionsby the Health and Safety Executive (HSE). During 1987and 1988, a study of some 2,000 site inspections revealeda worrying picture – with one-third of site agents andsupervisors being rated as having inadequate knowledgeof basic health and safety requirements. Importantly,during the period of the campaign, there was no decreasein the number of deaths or serious injuries (HSE, 1988).In essence, there are three elements to behaving safely:first, the knowledge of how to operate safely; second, (if necessary) the equipment to operate safely; and, third, themotivation to operate safely. Psychological interventionsattempt to improve safety performance by addressing thelatter. These interventions have taken several forms.Typically, posters and informational safety campaignsdesigned to improve safety through increased safetyconsciousness have not been consistently successful (e.g.Wilson, 1989; Saarela et al., 1989). Incentives have beenused successfully to improve safety behaviour (Peters,1991) but can be expensive. Further, they are apt to besuccessful only in the short term, as they do notnecessarily encourage the internalization of safe attitudesthat lead to long-term improvements regardless of material reward. In addition, they are apt to discourageoperatives from reporting accidents and near misses.The use of disciplinary action has shown limited success,largely for the reason that punishment is consistentlyheld to be less effective than positive reinforcement(Skinner, 1953). Punishment tends not to be effective if itis infrequent, delayed and of mild intensity. While manyargue that safety and productivity can go together ( and tidy access routes, speed, mobility), it is oftenheld that corners have to be cut if productivity is to bemaximized – leading to conflicting rewards. Summing upthe issue of conflicting reinforcers, Peters (1991) says:“when conditions are such that an unsafe act regularlyresults in…immediate positive reinforcement and anypotential punishing events are irregular, delayed andgenerally not very intense, people are apt to get hurt” (p.56). Another problem is that foremen are often reluctantto use punishment because of fear of resentment whichmay lead to lowered morale and lack of co-operation andproductivity (Peters, 1991). (Construction workers are notunknown to be more direct in their resentment –especially to the younger management who typically dealwith them on a day-to-day basis.)Thankfully, it is true that in the vast majority of cases anunsafe act does not result in an accident. Heinrich (1959)estimated that there is only one serious injury for every330 unsafe acts. Recent research (HSE, 1993) shows thatthere are 441 non-serious injury accidents for everyserious incident. Since serious accidents are relativelyrare, and are thus a poor measure of safety performance,it makes sense that monitoring and control be focused onthe behaviour which leads to them. Behavioural approaches Previous research (Shimmin et al., 1981) has shown thatin a sample of accident victims, two-thirds considered 5IMPROVING SAFETY BEHAVIOUR Improving safety behaviourusing goal setting and feedback T.W. Marsh, I.T. Robertson, A.R. Duff, R.A. Phillips, M.D. Cooper and A. Weyman  A description of the development and effects of behaviourally-based management techniques inimproving construction site safety Leadership & Organization Development Journal, Vol. 16 No. 1, 1995, pp. 5-12© MCBUniversity Press Limited, 0143-7739 Work outlined in the first section of this paper has beenreported in extended form in Duff  et al. (1994). This work wasfunded by the UK’s Health and Safety Executive. The content of this article, including any opinions expressed, are those of theauthors alone and do not necessarily reflect HSE policy.  their accidents to have been avoidable and caused byinappropriate behaviour or equipment usage. Thissuggests that, in the views of the workers themselves,something can be done to reduce accidents. Heinrich(1959) estimated that 85 per cent of accidents can beattributed to unsafe acts. It is, therefore, apparent that onemethod of accident reduction could involve the use of information feedback and goal-setting systems toinfluence the specific behaviours involved in accidents.This is exactly the approach at the core of this research. Goal setting  The literature on goal setting, as a procedure formanaging behaviour, is substantial. Wood et al. (1987), forexample, reviewed nearly 200 empirical studies. Goal-setting theory hypothesizes that goals are the immediate,though not sole, regulators of human action and thatperformance will improve when goals are hard, specificand accepted by the individual. Goal setting is held toaffect performance by: directing the attention and actionsof the individual/group; mobilizing effort; and increasingmotivation. In general, the literature on goal settingsupports these propositions and provides clear guidanceon how to operate the theory to good effect (Locke andLatham, 1990). Feedback  The research literature on the role of feedback indetermining performance effectiveness is clear inindicating the positive effect of knowledge of the resultsof one’s behaviour. Reviews of the research on feedback (e.g. Algera, 1990) demonstrate that performance isenhanced when management provides clear feedback of performance-related information. Goal setting, feedback and safety  These techniques for modifying behaviour have alreadybeen shown to be of value in safety. McAfee and Winn(1989), for example, showed that safety behaviour can beimproved by systematically monitoring safety-relatedbehaviour and providing feedback in conjunction withgoal setting and/or training. Chhokar and Wallin (1984)demonstrated how safety performance with feedback andgoal setting was better than with only goal setting in astudy of US metal fabrication workers. Reber and Wallin(1984) found similar results in a study of machinemanufacturers. Encouraging results have also beenachieved in construction-related industries in othercountries (e.g. Mattila and Hyodynmaa, 1988). Noprevious attempt, however, had been made to apply thetechniques in the UK construction industry. UMIST/HSE research – phase one In this phase, 13 building sites in the North West of England were studied for a period of one year. In order totest the effectiveness of the goal-setting and feedback techniques, we developed an accurate and reliablemeasure of safety performance which could be usedbefore, during and after their application. This wasachieved by identifying contributory behaviours in thechain of events which lead to accidents. In order toproduce a comprehensive list of unsafe behaviours, adetailed literature review of construction journals, HSEpublications, construction safety manuals and accidentrecords was undertaken. From an analysis of fatal andmajor injuries reported to the Health and SafetyExecutive (1988), 99 of these items were selected andincorporated in a questionnaire, which formed the basisfor a survey ( n  = 194) of construction personnel todetermine the perceived importance, or risk level, of eachitem in accident causation. Respondents rated each itemon three scales – frequency of occurrence, likelihood of aresulting accident and probable severity of such anaccident – and these dimensions were combined todetermine the relative importance of these items. Twenty-four items in total were selected and incorporated in asafety measure, which was used to evaluate the safetyperformance of construction sites in the experiments tochange safety behaviour. Items were categorized to formfour composite measures of safety: scaffolding; access toheights; housekeeping and personal protective equipment(PPE). This last was used only as a control.Previous research involving measures of safetyperformance had normally utilized an “all-or-nothing”(AON) measure of safe performance, i.e. either 100 percent safe or 100 per cent unsafe (Komaki et al., 1978).However, it is argued that a “proportional rating scale”(PRS) is better suited to cope with the constructionindustry environment. For example, a scaffold with only75 per cent of the required toe-boards correctly fixedwould be assessed as 25 per cent unsafe on that particularmeasurement item, rather than just “unsafe” – thuspermitting changes in safety performance to bemeasured with increased sensitivity. Further details of the advantages of this proportional scale are given inCooper et al. (1993). It is the far greater sensitivity of thismeasure over “all-or-nothing” measures which mostdistinguished this intervention from those of Komaki et al. (1978), Mattila and Hyodynmaa (1988) and others thatprecede it. An 11-point rating scale was employed foreach item – with a “not seen” option used to recordoccasions when there was no evidence of the item in eithersafe or unsafe condition (e.g. if none of the operativesobserved needed to be wearing eye protection). The scalewas anchored using expressions of amount which havebeen validated by previous research undertaken by Bass et al. (1974). Scores for each major category are calculatedby combining scores for the individual items observed.Goal-setting scripts were devised that included anexplanation of who the experimenters were, how the 6LEADERSHIP & ORGANIZATION DEVELOPMENT JOURNAL 16,1  safety performance levels were determined, what thecurrent levels of safety performance of the particularintervention category were and how frequentlyperformance feedback would be given. Charts capable of showing 42 weeks of measures, were designed andmounted in locations visible to all site personnel (e.g. onthe wall of the operatives’ canteen or on a fence at the siteentrance). The vertical axis, graded from 0 to 100 percent, indicated safety performance, while the horizontalaxis indicated the week number. Three coloured tapeswere used to plot safety performance levels on each chart.In addition, three coloured tapes with “target level”printed on them were used to indicate the goal to bereached.Goal-setting meetings were held in site canteens withgroups of 30-40 operatives at a time. On some large sites,more than one meeting was necessary to ensure theattendance of all operatives. Commonly, questionsregarding the provision of resources to allow and supportsafe systems of working, and the site management’s role,were raised. There was no attempt to either blame orreward individuals for their performance. The approachwas to consider the site and its personnel (operatives andmanagement) as a unit, and to give the feedback to allpersonnel, the philosophy being that it is theresponsibility of everyone to ensure safe workingpractices and conditions.An initial period of up to 12 weeks of data collection wasundertaken to dissipate any effects caused by thepresence of observers on site. After this, the procedureinvolved measuring safety performance for the threecategories for eight weeks to provide a base-line measure,following which the intervention was introduced. It isimportant to note that interventions were never aimed atthe control category (PPE). A “double-blind”methodology was used with independent observersemployed to conduct the safety audits. The observers,who had no knowledge of the research objectives, weretrained for a period of four weeks in the use of the safetyperformance measure. Observations on each site weretaken three times a week at differing times of the day andon different days of the week, to overcome any systematictime of day or week effects and to avoid operative andmanagement anticipation of the visits.Observation sessions took approximately 45 minutes oneach site, with the day and time of these observationssystematically varied. Each observer was allocated tospecific sites throughout the duration of the research.Consensual drift (i.e. where observers appear to drift fromthe srcinal consensus about the items, systematicallyand together, across observations) and inter-observerreliability were monitored regularly by random checkingof observations by a researcher who would independentlycomplete a measure for the same site at the same time.Using the percentage agreement method (Komaki et al., 1978) these checks showed 92 per cent agreementbetween observers and researchers. Findings – phase one Results, reported in detail in Duff  et al. (1994), showed asignificant increase in safety performance for the threeexperimental categories, but no such increase for thecontrol category. As an illustration, percentage scores forthe scaffolding category are shown in Table I.It was possible to draw the following conclusions fromthis phase of the project: q Safety behaviour can be objectively and reliablymeasured without excessive use of managerial orsupervisory resources, producing performancedata which can be used in many different safetymanagement strategies. q Goal setting and feedback can be used to producesignificant improvements in safety performance,at least over a period of several months. q Commitment of site management appears toenhance the effectiveness of the goal-setting andfeedback approach.It is clear that the methods of safety improvement usedduring this research had significant impact. These resultstherefore add to the findings of such as Mattila andHvodynmaa (1988) in suggesting that the use of goal-setting and feedback techniques can significantlyimprove safety behaviour on building sites. Theycontrast with the minimal effects that previous researchhas recorded for other interventions, such as informationsafety campaigns and safety training (Hale, 1984; Saarela et al., 1989). 7IMPROVING SAFETY BEHAVIOUR Table I. Summary of phase one results (category – scaffolding) BaselinePost-interventionSitemean (sd)mean (sd)Plastic processing factory79.80 (3.33)92.80 (2.35)New hospital wards71.87 (6.82)77.27 (2.24)Large office and retaildevelopment93.80 (2.90)93.17 (1.80)City centre apartments59.27 (6.28)73.83 (3.46)Offices in city centre69.64 (9.20)70.50 (7.54)Bank premises in city centre87.00 (7.73)92.67 (2.61)  Research – phase two At the start of 1993 a further two-year study wascommissioned by the HSE for UMIST to continueresearch in this area. Specifically, research was intendedto focus on the following: q Management commitment. Although no attemptwas made to evaluate the effect of managementcommitment as an integral part of phase one of theresearch, it became apparent that this wasprobably having a vital impact on some of theresults. The two best-performing sites overall werethose where management attended all themeetings with operatives at the commencement of the intervention. These findings are also broadly inline with an overview of a variety of managerialinterventions by Rodgers et al. (1993), who foundthat “studies have consistently reported thatcommitment from top management is essential”for an intervention to succeed (p. 151). q Application by company personnel. If thisintervention is to be widely adopted it will need tobe implemented by company personnelthemselves. It was felt likely that this could lead toa polarization of performance. In “good”companies, where management commitment isstrong and support and encouragement are givento facilitators, the intervention should be evenmore successful than when facilitated byoutsiders. However, in companies wheremanagement commitment is low – and facilitatorsare given little support – we might expect theintervention to be relatively less successful. Methodology  Eighteen sites have been actively involved in research todate. Of these, seven are completed, and in 11 we are stillcollecting data. A further six sites are due to becomeinvolved in the next few months. Sites have variedgreatly in nature, from civil engineering to building andfrom small (15 operatives) to large (more than 200operatives).The methods used in phase two are identical to those inphase one, except that the measures are taken, the goal-setting sessions facilitated and the charts updated byorganizational personnel after training from the researchteam. Observers are given a half-day training session inthe theory of goal setting and feedback in a classroomsetting. During this formal input, observers are alsointroduced to the measure and given practice at scoringexample data. A further half-day of training in the use of the measure is given on each observer’s own site. Earlyfindings suggest that this one-day training programmeis 1ong enough to train observers to the necessarystandard. Findings – phase two Results are still being collected and have not yet beenfully analysed. Results to date, however, have beenpositive in the great majority of cases, for all fourcategories. Two typical examples are shown in Figures 1and 2. Many of the issues raised below are still beingstudied and the conclusions discussed are necessarilytentative. Management commitment  As expected, the role of management commitmentappears absolutely vital. It clearly has a high impact onall aspects of the intervention, and will be referred toduring each of the sections below. The observer(s) Not surprisingly, the role of the observer seems to be of vital importance. Two factors are particularly interesting.First, there is the psychological make-up of the observer.Second, the position within the organization that the 8LEADERSHIP & ORGANIZATION DEVELOPMENT JOURNAL 16,1 10090807060501 4 5 10 15 Mean qqqqqqqqqqqqq GoalMean Goal-settingBaseline Week number    P  e  r  c  e  n   t  a  g  e  s  a   f  e   t  y  p  e  r   f  o  r  m  a  n  c  e q   Figure 2. Site two (site finished). Housing refurbishment – North-East England (category – scaffolding) 100908070605015 7 8 10 15 20 25 27 30 Mean qqqqqqqqqqqq qqqqqqqqqqqqq q q GoalMean Goal-settingBaseline Week number    P  e  r  c  e  n   t  a  g  e  s  a   f  e   t  y  p  e  r   f  o  r  m  a  n  c  e   Figure 1. Site one (site continuing). Law courts North-West England (category – scaffolding)
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!