between organizational factors and safety and contingency measures on offshore petroleum platforms.

RUNDMO T. Associations between and and measures on offshorepetroleumplatforms. OBJECTIVES - The study determined the associationsbetween organizationalfactors and safety and contingencymeasuresamongemployeeson the Norwegian continentalshelf. METHODS - A self-administeredsurvey among offshore petroleum personnel was conducted in the spring of 1990. The survey drew respondents from five companies and eight installations. The re sponserate was 92% (N = 915). RESULTS - Substantialcorrelations werefoundbetweenmanagementand employeecommitmentand involvement in safety work, social support, attitudes towards safety measures, and the personnel's satisfaction-dissatisfaction with the safety andcontingencymeasures. CONCLUSIONS - Employeeperceptionsof greater managementcommitment,social support, and sub jective evaluations of priorities of safety versus production goals were all important predictor vari ables for employeesatisfactionwith safetyand contingencymeasures.

sions: safety traimng programs, management attitudes towards safety work, effects of safe conduct on promotion and social status, the level of risk at the place of work, priorities of work tempo versus safety, and the status of safety officers and safety committees. Brown & Holmes (7) reduced the model to a three-factor model consisting of management concerns, management safety activities, and employee risk perception. Dedobbeleer & Beland (9) tested Brown & Holmes' safety climate model among workers employed on construction sites. They reduced it to a two-factor model consisting of management commitment (management concern and management safety activities) and worker involvement.
Organizational and social factors affecting safety prevention work must be distinguished from safety and contingency measures. Organizational factors can affect the safety work and therefore also satisfaction-dissatisfaction with the safety and contingency conditions among employees. If so, it is reasonable to expect that organizational factors affect variations in satisfaction-dissatisfaction with safety and contingency factors.
The specific objective of this paper was to determine the effect of organizational and social factors (ie, management and employee commitment and involvement in safety work, social support, and employee attitudes towards safety prevention work) on satisfaction-dissatisfaction with safety and contingency measures on offshore petroleum platforms.

Subjects and methods
Sampl e A sample, intend ed to be representative of the offshore petroleum industry on the Norwegian Continental Shelf, was selected. It was taken from eight platforms and five companies. All personnel present on the platforms at a specific time took part in the survey. Out of those living on the platform s at the time, 92% replied to the questionnaire (N =915).

Measures
In the quest ionnaire the respondents were asked to assess their satisfaction with the safety and contingency factors on the platform, manag ement and employee commitment, and involvement in safety work, social support from management, supervisors and colleagues, and their own attitudes toward s safety measures.
Safety and contingency fa ctors. Safety and contingency aspects included employee evaluation of protective measures, safety instructions, and training . The respondents were asked to rate their own satisfaction-dissatisfaction with the following items: control and inspection in safety work, work instruction s, safety instructions, follow -up measures taken after injuries and accidents, first-aid training, contingency training, safety training, order and cleanliness at the place of work, access to emergency exits and escape route s, prote ction and safety device s on machines and equipment, markin g and sign posting, availability of personal protective equipment , and use of personal protective equipmen t. The evaluation included ratings on a five-point scale for each test item. The scale ranged from "very satisfied" to "very dissatisfied." Comm itment and involvement in safety work. The respondents were asked to rate to what extent they believed that the following people were concerned with their safety and in participating in accident prevention work : platform management, immediate supervisors, trade unions , fello w workers, and the respondent himself (10). For this factor, a five-point rating scale was also applied. The scale ranged from "very concerned" to " not concerned at all" in improving safety and involvement in accident prevention work.
Social support. The perception of instrumental support (ie, aid in the form of changes in the environment), emoti onal support (ie, feeling of trust and concern), and informal support (ie, advice, suggestions and information) was measured . The respondents were asked the followin g question s: "How Scand J Work Environ Health 1994, vol 20, no 2 much support do the following peopl e provide you with to manage your work?"; "To what extent can you talk with the follo wing people?"; and "How much do you feel you can trust the following people when things get difficult at your place of work ?" They were asked to give separate ratings for supervisors, fellow workers , and trade union repre sentatives in reply to each of the three question s. The five-point scale which was used ranged from "very much support" to "very little support." Social support has been measured in the same way in previously published studies ( I, 2).
Attitudes towards accident pr evention work. Selected items were applied from a questionnaire developed for the evaluation of safety attitudes in industry (6) . The following items were included: "sometimes it is necessary to depart from safety requirements for the sake of production"; "good operational econom y is often in confli ct with measures to improve personal safety"; "rules and instructions relatin g to personal safety sometimes make it difficult to keep up with production targets"; "sometimes it is necessary to take risks to get a job done"; "whene ver I see safety instructi ons not being complied with , I call attention to it on the spot"; "many minor inj uries and minor accidents are an indic ation that serious accidents can also easily occur" ; "safety measures only shift the danger from one area to another" ; "occupational accidents are often the result of bad planning and poor management"; "calling attention to breaches of safety can easily be felt as unnecessary hassle"; "good proposals on how to improve safety are often dropped if they cost too much"; and "many accidents ju st happen , there is littl e one can do to avoid them." The five-point scale used to measure attitudes , ranged from "agree strongly" to "disagree strongly."

Analysis
Because of the great number of single items used in the measurement, it was first con sidered necessary to replace single test items by indice s depicting the underlying structure of the evaluations (belief dimen sions). To detect the underlying factor , principal component analysis with iteration and varimax rotation , which yields orthogonal factors, was used. Missing values were exclud ed listwise (ie, all of the cases with missing values for one or more of the items were eliminated). For testing the reliabil ity of the dimensions Cronbach ' s alpha was used For each person the scores on the test item s belonging to a certain index were then added without differential weighting and a score was determined for each dimen sion. For testing the differen ces in sig-nificance due to personnel group and employment conditions the chi-square test, t-test, and oneway analysis of variance were used.
The main objective of the study was to predict how management and employee commitment and involvement, attitudes towards safety measures, and social support affected the employee's evaluation of the safety and contingency measures. LISREL (analysis of linear structural relationships) was applied to measure the contribution of these predictors to the endogenous variables (satisfaction-dissatisfaction with the safety and contingency measures) (12,13). The same statistical procedures have previously been described in greater detail in a study of the relations between safety and contingency factors, risk perception, and accidents (14).
LISREL is well suited for estimating relations between latent or unobsei ved variables, in which case each has one or more indicators (test items) being measured. The method is also appropriate for analyzing the relations between one or more directly measured endogenous variables and one or more unobserved variables, what is called MIMIC (multiple indicators and multiple causes) models, as well as ordinary path analysis. LISREL was applied to test the MIMIC models.
The maximum likelihood method (LISREL-ML) was used. This method proved to be appropriate for the data. Correlations between the models and the data were examined by means of the chi-square test and tests of general adaptability (goodness-of-fit index) and modified general adaptability (adjusted goodness-of-fit index). The models presented in this paper all satisfied the demand for nonsignificance (P>O.IO).

Dimensions of the safety and contingency measures
Factor analysis identified the following three dimensions for the safety and contingency measures: protective measures, instruction, and training. Table I presents the factor loadings for the test items belonging to each of the dimensions. The eigenvalue was greater than 1.0 for all of the factors. The structure that emerged when all the respondents were included in the analysis was used as an a priori factor structure for each personnel group. The groups were administration and management personnel, production personnel, drilling personnel, maintenance and construction personnel, technical-mechanical personnel, and catering personnel. Reliability was tested separately for each group. As seen in the table, Cronbach's alpha was found to be satisfactory for all of the personnel groups. Table 2 shows the percentages "not satisfied" with the three dimensions of the safety and contingency measures in relation to personnel category. The category "not satisfied" contained those who were strongly dissatisfied, dissatisfied, and neutral (15). The table indicates the following: (i) the greatest percentage of persons were "not satisfied" with training (contingency, safety, first-aid), (ii) many were "not satisfied" with work instructions and safety instructions, and (iii) protective measures represented the factor producing the greatest satisfaction.

Direct evaluations of the safety and contingency measures
To test whether the differences in various groups of personnel were significant, a oneway analysis of

Organizational and social factors affecting satisf action-diss atisf action with the safet y and contingency measures
The next step was to determine the contribution of (i) management a nd e m ployee com mitme nt a nd involve me nt in sa fe ty work, (ii) att itudes to ward s preve ntive measures, and (iii) so cial suppo rt to sa tisfaction-dissati sfact ion with the safe ty and co ntingency measu re s. In order to find efficient pre venti ve mea sures, kno wledge was needed about the way these factors affec ted the personnel' s evalua tio n of th e dependent fac tors. The reli ab ility of the dimen sion s commitment and involvement in safety work was fo und to be at a satisfactory le vel. Cronbach's alpha was 0 .825. The sa me was true for social support, for whi ch Cronbach's alpha was 0.8 55 .
Factor a na lysis detected the fo llowing two di me nsio ns of attitudes towards sa fe ty work: (i) pri orities of safety ve rs us producti on goals and (ii) attribution of acc ide nt causes. Table 3 o n page 126 presen ts the fac tor load ings fo r each of the tes t items. Th e re liability of the fac tors was tested se pa ra tely fo r each personnel group. As see n in the tabl e, Cronba ch ' s alph a was found to be satisfac tory for ea ch group for th e respondents' priority of sa fety versus product ion goals. Due to lac k of reliability, the attribution of accident causes was excluded fro m further ana lyses .
LISREL was used to es ti ma te th e co rre lations between th e va ria b les . Th e mod el presented in figure I con sists of a si ng le unobser ved e ndogeno us latent 125 Scand J Work Environ Health 1994, vol 20, no 2 Table 3. Dim ensio ns of t he attitud es tow ards safety work and the reliabili ty of the factors. iables we re directl y me asured b y three x va ria bles (X = XI' X 2' x.), which affect 11 directl y, (X =~). (T he X variab les were co m mi tme nt and inv olve me nt in safet y work, pr iority of sa fe ty versu s producti on goals, a nd social support). Single test items were replaced by indices in the mod el. Th e ana lysis re veal ed th at management a nd e mpl oye e co m m itment and in vol vement in sa fety work, attitudes towards safety measures, and social suppo rt co ntri bute d sig nifica ntly to sa tisfactio n-d issatisfaction w ith the safety and co nti ngenc y mea su res [coefficient of determination (R2) = 0.24]. Contribution s from th e three ex oge nous fa ctor s, X I' x 2 ' x., were R 2 x 100 (ie, 24 %). In addition it was fo und th at XI' man agem ent and employ ee co mmi tme nt and in vol vement, was the exogenous fac to r cont ributing most to satisfactio n-d issatisfactio n (Y I I = 0. 36 ). Pri ori ty of sa fety ve rs us production goals and soc ial suppo rt also contributed s ig nificantly. Fi nally th e corre latio n between co m m itme nt and in volvement and social support was 0.32. Th ere was a lso a subst antial correlation bet ween commitment and inv o lv ement and priority of safe ty vers us production goals.
Nex t, se parate a nalyses were conducted for each personnel g ro up, as well as for each of th e employmen t co ndi tio ns, a nd th e effects o n the e ndogeno us factor were com pared. Se ven analyses were carried out. (See table 4. ) The results can be sum m arize d as follows : (i ) the mod el (figure I) was fo und to be suitable for th e data unde r all of th e specified co nd itio ns and it satisfie d th e demand fo r nonsi gn ificance in all th e dat a se ts; (ii) th e model was th e most successful in e xpl aining sa tisfactio n-d issatisfac tio n for catering personnel (R 2 =0.33), for maint enan ce and co nstruction personnel (R 2 =0.30), and for dr ill ers (R2 = 0.27) but was not so suc cessful in e xp laining the ev aluation am ong technical-mechan ical pe rsonnel (R 2 = 0 .1 3); (iii) manage ment a nd emp loyee co m mi tme nt and in vol vement was the most im portant pr edi ct or.

Discussion
The pe rsonnel' s assessments of management a nd employe e com m itme nt a nd involvement in sa fety work, atti tudes to ward s accident pre ve ntion , an d soc ia l su ppo rt correla te d sig nificantly wi th th eir satisfac tio n-d iss atisfactio n w ith the sa fe ty and co nti nge ncy fac tors. Th e are as which had the strongest effec ts were (i) co m mi tment and in volvement in safety work by supervisors and management (o verall this factor had th e most important effect on the e valuation of safe ty and contingen c y me asu res) (i i) socia l support (manageme nt and s upe rvisor sup po rt were particularly sing led out as im porta nt); and (ii i) att itudes towards accide nt pr e vention wo rk (a ttitudes provided a s ma ller contribution th an th e other two dimen sion s with regard to variati on in sat isfaction-d issatisfaction with safety and continge nc y measures).
Scand J Work Environ Health 1994, vol 20, no 2 T he mo st efficient way to reduce d issati sfaction w ith the safety and conting e ncy facto rs lies probab ly in steps ai me d dire ctl y at impro vin g work and safety instructi ons, training and sa fe ty devi ces, and steps to impro ve organizational factors affecting safety and contingency measures (ie, management and supe rv isor commitment a nd involveme nt). Social s up po rt a nd atti tudes to ward s accident pr e vention work a lso co nt ribute d co ns idera bly to the va ria tio ns in satisfac tion-dissa tisfaction. In add itio n, s ubstantial co rre la tio ns were fo und bet we en management and supervisor commitment and involve me nt and so ci al support. Efforts sho uld therefore be di rected at improving management and supe rv iso r co m m itment and in vol vements and social support.