a safety climate measurement for construction sites

Jourml ofSa& Research Vol. 22, pp. 97-103.1991 Q 1991 National Safety Council and Pergamon Press plc 0022-4375/91$3.00 + .OO Rited in the USA A Safety Climate Measure for Construction Sites Nicole Dedobbeleer and FranCok Beland This study tests Brown and Holmes’ (1986) three-factor safety climate model on construction workers. In this model, climate was viewed as molar perceptions people have of their work settings. Data were collected by a selfadministered questionnaire in a cross-sectional
of 7
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.
Related Documents
  Jourml ofSa& Research Vol. 22, pp. 97-103.1991Q 1991 National Safety Council and Pergamon Press plc 0022-4375/91$3.00 + .OORited in the USA A Safety Climate Measurefor Construction Sites Nicole Dedobbeleer and FranCok Beland This study tests Brown and Holmes’ (1986) three-factor safety climate model on construction workers. In this model, climate was viewed as molarperceptions people have of their work settings. Data were collected by a self-administered questionnaire in a cross-sectional survey conducted among 384workers employed in nine nonresidential construction sites in Baltimore, MD. The response rate was 71%. Results using two linear structural relations(LISREL) procedures (maximum likelihood used by Brown and Holmes andweighted least squares) indicated a good model fit. The weighted leastsquares procedure, which is more appropriate for our data, revealed that atwo-factor model provided an overall better fit. The two factors were (a)management’s commitment to safety and (b) workers’ involvement in safety.This model emphasizes management and workers’ involvement in safetymatters. Results also suggest the necessity of addressing concerns of these twogroups in safety policies.Although the concept of climate hasappeared in the literature for over 20 years,disputes still rage about what it means andhow it is measured, In industrial organizations,safety climate models have been constructedfor production workers by Zohar (1980) andBrown and Holmes (1986). In these two mod-els, climate was viewed as molar perceptionspeople have of their work settings. FollowingRyan (1970), and Dieterly and Schneider(1974), these models assumed that these per-ceptions are developed because they are neces-sary as a frame of reference for gauging the Nicole Dedobbeleer, Sc.D., is an Assistant Professor,Department of Preventive and Social Medicine and memberof the Inte.rdisciplinary Research Group in Health (GRIS),Universitk de Montreal, Montreal, Canada.Francois Beland, Ph.D., is an Associate Professor.Deparkent of Health Admmistration and member of the(GRIS), Universid de Montreal, Montreal, Canada.Portions of this paper have been presented at theInternational Conference on Strategies for OccupationalPrevention, Stockholm, Sweden, 21-22 September 1989.Summer 1991Molume 22lNunrber 2 appropriateness of behavior. Based on a vari-ety of cues present in their work environment,employees were believed to develop coherentsets of perceptions and expectations regardingbehavior-outcome contingencies and behavingaccordingly (Frederiksen, Jensen, & Beaton,1972, and Schneider, 1975a, 1975b). As inboth models, safety climate was viewed as anindividual attribute as opposed to an organiza-tional attribute, the term psychological climatewas applied to the two models (James &Jones, 1974).According to James and Jones, psychologi-cal climate is subject to serious theoreticaland methodological questions. With respectto safety climate, these questions may beraised about the number of dimensions rele-vant to the safety climate domain. Zohar’smodel, constructed on an Israeli population,included eight dimensions: (a) importance ofsafety training programs; (b) managementattitudes toward safety; (c) effects of safeconduct on promotion; (d) level of risk at 97  workplace; (e) effects of required work paceon safety; (f) status of safety officer; (g)effects of safe conduct on social status; and(h) status of safety committee. This modelwas validated by Brown and Holmes on anAmerican sample of production workers andwas reduced from an eight-factor to a three-factor model. In this three-factor model, thefollowing dimensions were retained: (a)employee perception of how concerned man-agement is with their well-being; (b) employ-ee perception of how active management is inresponding to this concern; and (c) employeephysical risk perception.As the validity of a safety climate modeldepends in part on its applicability to groupsother than production workers, we testedBrown and Holmes’ three-factor model onconstruction workers. To perform this test, weused two (LISREL) procedures: (a) the maxi-mum likelihood method chosen by Brown andHolmes, and (b) the weighted least squaresmethod, which is now available and is moreappropriate for polychotomous ordinal levelvariables. We will compare the resultsobtained with these two procedures.h4ETHODSSetting and SubjectsThe study was conducted on nine nonresi-dential construction sites located in theBaltimore, MD, Metropolitan Area. Thesesites were selected from Reports onMetropolitan Building Permit Activity devel-oped by the Regional Planning Councilaccording to the following criteria: (a) con-struction ranging from the framing stage to thestage prior to the finishing stage; (b) new pro-jects or new projects with alterations: and (c)projects valued at $500,000 or more. Fifteencontractors had construction projects conform-ing to these criteria from August to December1983, all were contacted. Seven contractorsrefused to collaborate (i.e., problem with insur-ance coverage for observers, fear of work dis-ruption, and labor time costs induced by theadministration of the questionnaire), eightagreed, and two were found ineligible (i.e., sitedid not correspond to criteria). Among the sixremaining contractors, three had two sites andthree had one site, for a total of nine sites. 98 The study population was a sample of 384eligible workers drawn from the 572 workersemployed at the nine sites. Workers were eligi-ble if they were allowed to participate in thestudy by their employer and if at least 50% ofthe safety practices examined were applicableto their work. These safety practices were asampling of the Occupational Safety and HealthAdministration (OSHA) safety standards relat-ed to ladder use, scaffold use, and personal pro-tective device use (i.e., hard hats, safety shoes,and safety belts). The survey response rate was71%, or 272 workers. Eighteen workers wereexcluded because of response omission to themajority of the questions, particularly to ques-tions on safety practices.Instruments and Data CollectionA self-administered questionnaire was usedto collect data from workers at the worksite.The closed-format questionnaire incorporateditems on socio-demographic and employmentcharacteristics, safety practices, safety train-ing, safety instructions at initial employment,knowledge about safety practices, attitudestoward safety practices, perceptions of theworkplace, and other social-psychologicalfactors. The researcher was present while therespondents filled out the questionnaire at sixsites. A detailed consent form was given tothe workers before they started the question-naire to insure that they fully understood thenature of their participation in the study. Measures The three dimensions of Brown andHolmes’ safety climate model were measuredwith nine variables. These variables were cho-sen to represent safety concerns in the con-struction industry and were not all identical tothose included in Zohar’s questionnaire. Eachdimension included three variables. Appendix1 presents the questions used to measure thenine variables.Employee perception of management con-cerns was based on items indicating employ-ee’s perceptions of management’s attitudetoward safety practices, and employee’s per-ceptions of management’s attitude towardworkers’ safety and foremen’s behavior.Management’s attitude toward safety practices(MANA. PRACTICES) was derived from theworker’s perception of the importance of safe- Journal of Safety Research  FIGURE 1 TWO MAXIMUM LIKELIHOOD SOLUTIONS FOR THE SAFETY CLIMATEMODEL IN CONSTRUCTION INDUSTRY A) Three-factor model (total coefficient of determination: .884) Factors:.46 Management concerns Indicators: Errors:Practices Safety ForemanInstructions Mectings Equip.48 1 .14 1 .61 t .I9t .L8j .34 1elez e3e4e5e6e7e8 e9 B) Two-factor model (Total coefficient of determination: 850).63 Factors: Management 5 Workers commitmentinvolvement Indicators: Practices Safety Foreman Instructions Equip Errors: 5 e2 e3 e4e5 e6e7e8 e9 (1) The measure of association shown on the mows between the factors are correlation coerficients(2) The measure of association between the factors and their indicators are regression coefficients(3) The measures of association between the indicator: and the ‘e’ can be interpreted as the error of prediction in regression equations Summer 1991lVolwne 22INumber 299  ty practices to top management and was mea-sured by a four-point rating scale rangingfrom “very important” to “not at all impor-tant.” Management’s attitude toward workers’safety (MANA. SAFETY) was derived fromthe worker’s perception of this attitude andwas measured by an itemized rating scale.Foreman’s behavior (FOREMAN) wasderived from the worker’s perception of theactions taken by the foreman to enforce safety,and was measured by an itemized rating scale.Employee perception of management safetyactivities included safety instructions, safetymeetings, and employee’s perceptions of theavailability of proper equipment. Safetyinstructions (INSTRUCTIONS) referred tothe employee’s self-reported exposure toinstructions on the safety policy of the compa-ny at the time of initial employment. Safetymeetings (MEETINGS) was the worker’s per-ception of the presence of regular safety meet-ings at the worksite. Proper equipment(EQUIP) was the worker’s perception of theavailability of proper equipment at the work-site, and was measured by a five-point ratingscale ranging from “always” to “never.”Employee’s physical risk perception includ-ed employee’s perceived control, perception ofrisk-taking, and perceived likelihood ofinjuries. Perceived control (CONTROL) wasdefined as the worker’s perception of controlover his/her own safety on the job, and mea-sured by a four-point rating scale. Perceptionof risk-taking (RISK) referred to the worker’sperception of the extent to which risk-taking ispart of the job, and was measured by a three-point rating scale ranging from “very much” to“not at all.” Perceived likelihood of injuries(INJURIES) was defined as the worker’s pcr-ception of his/her susceptibility to work-relat-ed injuries in the next 12 months, and wasmeasured by a four-point rating scale rangingfrom “very likely” to “not at all likely.”AnalysisTo test for the replication of Brown andHolmes’ three-factor structures in our data,two LISREL procedures were applied to thenine variables described above. The maximumlikelihood method chosen by Brown andHolmes was used first. Our second step was touse the weighted least squares method to esti-mate the parameters of the model (Joreskogand Sorbom, 1988), as the data analyzed inthis study are made up of polychotomousordinal level data. This method is more appro-priate for our data than the maximum likeli-hood method, which is recommended for nor-mally distributed and continuous variables.Because we had ordinal variables, we alsoentered polychoric correlations in the LISRELmodel. In the second step, the three-factormodel was tested against a two-factor modeland a one-factor model using the X2 statistics.RESULTSResults from the maximum likelihoodmethod indicated that Brown and Holmes’model was supported by our data (X2 = 28.67, df = 24, p = .233). Further tests of this modelwere then conducted by reducing the numberof factors from three to two (Figure 1, PanelA and B for a description of the models), andthen from two to one. The two-factor modelwas barely acceptable (X2 = 36.57, df = 26, p = 0.082). The one factor model was rejectedaccording to the X2 criterion (X2 = 51.07, df =27, p = 0.003). The difference between thethree-factor and two-factor models was thentested. Results indicated that there was a sta-tistically significant difference between thetwo models (AX2 = 7.90, Adf = 2, p = 0.019).Subsequently, Brown and Holmes’ safety cli-mate model was retained as the best fit model. 100 TABLE 1TEST OF THE HOLMES AND BROWN’S SAFETY CLIMATE MODEL USING THEWEIGHTED LEAST SQUARES METHODModel df X2 pValues X2 df p Values1 factor 27 43.65 0.0222 factor 26 21.19 0.732 22.46 1 O.ooO3 factor 24 16.45 0.871 4.74 2 0.093 Journal of Safety Research
Related Search
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