Obras de construcción: conectando predictores de la estructura organizativa con el nivel de riesgo

Nuestra investigación estudia la relación entre la gestión organizativa y el nivel de riesgo (NR) en obras de construcción. Analizamos empíricamente los efectos en NR causados por cuatro áreas de gestión organizacional de la obra. Nuestra hipótesis general es que la gestión organizacional es un factor importante que afecta a las condiciones de seguridad. La revisión de la literatura aborda este tema y demuestra que hay una conexión importante pero que ha sido escasamente abordada. Para desarrollar nuestra investigación, hemos visitado y evaluado aproximadamente 1000 obras de construcción en España durante el periodo 2003-2009. Nuestra muestra está formada obras de edificación de diferente tipología, propiedad, grado de complejidad, medios, etapas de trabajo, organizaciones y niveles de seguridad. Mediante nuestra propia herramienta para la evaluación de obras, hemos analizado tanto aspectos de seguridad como organizativos. Por un aparte, hemos construido un índice de riesgo global (IRG) para la obra usando 11 variables de riesgo. Por otra parte, hemos obtenido datos e información para describir y caracterizar 10 variables organizativas para cada obra. Todas estas variables se clasificaron en una de las mencionadas cuatro áreas de gestión organizacional: complejidad de la obra, recursos de la estructura organizativa, complejidad del diseño organizativo y recursos para la gestión de la seguridad. Proponemos utilizar metodología SEM (modelos de ecuación estructural) para probar nuestras hipótesis relativas a relacionar niveles de gestión organizacional e IRG, así como validar nuestro modelo de medición de seguridad y gestión de construcciones. Una de nuestras aportaciones es la gran cantidad y calidad de los datos de campo obtenidos, poco común en la literatura. Otra contribución adicional es identificar los aspectos de la gestión que deben ser reforzados con el fin de prevenir los riesgos en la obra, convirtiendo la información en un enfoque proactivo.
Main Author: 
José M.
Universidad Illes Balears
Francisco José
Forteza Oliver
Universidad Illes balears
Sesé Abad
Universidad Illes Balears


Originally, most health and safety (H&S) research on construction began by highlighting the accident rates problem as well as the special nature of construction (Baxendale & Jones, 2000; Cheng, Leu, Lin, & Fan, 2010; Mahmoudi, Ghasemi, Mohammadfam, & Soleimani, 2014; Wu, Liu, Zhang, Skibniewski, & Wang, 2015). At the same time, it seems that there is something intrinsic in construction sector that produces these risks. Currently, risk level (RL) assessment research has evolved from an accident-based approach towards a more prospective and holistic one, characterised by technical analysis adding organization and human factors (Sgourou, Katsakiori, Goutsos, & Manatakis, 2010). Despite this tendency, most of the current research is still based on accidents (Hollnagel, 2008; Khanzode, Maiti, & Ray, 2012). Thus, there are a small number of studies where authors use precursor analysis as an alternative to classical accident approach. These authors criticize about reactive research techniques that use lagging indicators and they propose different leading indicators (predictors) to obtain information before an accident occurs (Grabowski, Ayyalasomayajula, Merrick, & McCafferty, 2007; Hinze, Thurman, & Wehle, 2013; Rozenfeld, Sacks, Rosenfeld, & Baum, 2010; Sparer & Dennerlein, 2013; Toellner, 2001). In this research we try to link organization variables (OV) with RL in order to propose them as another set of predictors of risk.

Organisational factors have arose as a relevant issue for risk construction sites research. Since Hoewijk (1988) proposed that the vertexes of the “Organization Triangle” formed by structure, culture and processes, are mutually dependent and conform workers behaviour, others researchers have analysed this organisational side of the problem of safety. For construction companies, adopting this approach is complex because the especial features of the sector as, for example, the temporary nature of sites, their physical distance from company headquarters, the low level of standardization of processes and so on (Wilson, 1989). Besides this, the sector is also characterized for the special conditions of agents structure, business processes and operational levels (Donaghy, 2009; HSE, 2009). For Swuste et al. (2012) construction companies are similar to an organic structure that manifests itself in its processes. Although process may determine the organisational structure on site, the resources of the head company are determinant to empower enough site resources.

There is a certain consensus about the relationship between organisational factors and safety performance (SP). In fact, a selection of nonspecific construction SP assessment methods was analysed by Sgourou et al. (2010) and all of them include organisational features. But a low number of field research, specifically on construction sites, have connected and concrete these relations (Ai Lin Teo & Yean Yng Ling, 2006; Fang, Huang, & Hinze, 2004; Mohamed, 1999; Wu et al., 2015) and even fewer have linked organisational and complexity with risk level assessed on site (Fang, Huang, et al., 2004).

Literature review and hypotheses

Most research try to find connections between different aspects of safety management systems and SP. We do not analyse safety management systems, because in our construction companies sector (manly small and medium size) they did not exist. Differently, we try to identify which organizational factors, related with complexity and resources, affects the safety on sites. We neither analysed SP in the present study, looking instead directly into the risk level. SP and prospective SP are concepts under investigation that very few empirical studies have analysed (Wu et al., 2015). According to Ghasemi, Mohammadfam, Soltanian, Mahmoudi, & Zarei (2015) SP has two aspects: safety compliance (e.g. working conditions, protections, procedures and rules) and safety participation (e.g. motivations, safety meetings participation). Therefore, looking directly into the risk level, we are restricting our study to the safety compliance aspect of SP, i.e. the working conditions on site. In sum, what we will try to link in this research are organizational factors and risk level. Next we will review the literature studying both dimensions and will propose our set of hypotheses.

Mohamed (1999) tried to link safety conditions with the effectiveness of safety management activities. The author developed a safety management index (SMI) based on six variables (policy, management declaration, meetings, internal audits training and awareness programs). Then he compared it with a safety performance index (SPI) based on six variables too (SP records, presence of subcontractors in safety discussions, planning with consideration of hazards, personnel’s rewards, appointment of trained safety officers and record of intoxications). Doing so he analysed the relationship between them through correlation analysis and cluster analysis. His results showed that no significant positive correlation between SMI variables and SPI variables could be detected.

Some other authors linked safety management practices with project execution performance (E. W. L. Cheng, Ryan, & Kelly, 2012). They concluded that safety committees are one of the most important issues in order to improve project performance. Other research found a reduction of accident ratios, more competitiveness and better economic performance in those firms having better safety management (Fernández-Muñiz, Montes-Peón, & Vázquez-Ordás, 2009).

Specifically, Fang, Huang, et al. (2004) developed a so called safety management index (SMI) based on seven organizational factors. They used regression analysis to explain a safety performance index (SPI) at construction sites with their SMI. The dependent variable (SPI) is derived from inspection records of physical safety conditions, satisfaction of site personnel and accident frequency. Their results showed that among the seven organizational factors previously proposed only organizational structure, economic investment, and relations between labour and management were significantly related to their SPI.

Fang, Xie, Huang, & Li (2004) identified eleven factors that correlate closely with onsite safety management performance. Using factor analysis these authors extracted out 11 common factors. Among them, five were prioritized: foreman, worker, crew, manager and safety training.

Wu et al. (2015) proposed that SP is linked to four different aspects: safety climate, culture, attitude and safety behaviour. They proposed measure SP based on historical evidence of accidents and safety records of inspections. This study has similarities to ours, because safety behaviour is measured through working conditions (safety working practices) like we do, as for example fall protections in height risk places, attention and measures of the lifting jib, safety measures in deep excavation, temporary electrical installation, safe use of scaffolds and emergency measures.

Törner & Pousette (2009) pointed out that any study of safety management must to develop a set of organizational measures. Those measures were elicited based on interviews to workers and managers, and they can be categorized in four main work safety preconditions:

“(1) Project characteristics and nature of the work, which set the limits of safety management; (2) Organization and structures, with the subcategories planning, work roles, procedures, and resources; (3) Collective values; and (4) Individual competence and attitudes.” (p.401)

Up to this point we have reported here the most relevant literature addressing the relationship among H&S issues and organizational factors, either theoretically or empirically. In the following section we develop our theoretical model relating risk in construction sites and organizational factors.

Theoretical model and hypotheses

Based partially on technical knowledge and the evidence found in the empirical literature we propose here that there may be a model that explains risk on construction sites as a function of two broad organizational factors, complexity and resources. Therefore, our proposal is that risk is a function of both factors as expression (1) reflects:

Risk=f(complexity;resources) (expression 1)

Our model of risk and its empirical test are presented here as one of the major contributions of our research.

Following we explain in more detail the composition of the four factors classification from literature review.

F1) Site complexity

Literature review shows that complexity is an important factor and relevant for our research. We understand complexity as the combination of three main variables: complexity of project, size of site and stage complexity.

Hon, Chan, & Wong (2010) linked special characteristics of accidents with type of project, considering works on repair, maintenance and extension works. However, they also found common factors of risk which are present in the entire construction industry. In conclusion, for these authors the type of work project created differences but does not exclude other common problems inherent to the construction sector. Hatipkarasulu (2010) categorized projects and identified high risk typologies with relation to mortal accidents. Residential and commercial typology present more accidents and falls from height are the principal cause. Manu et al. (2010) include all our variables in their mentioned CPF. The nature of project includes type of work (e.g. new work, repair/refurbishment/maintenance, and demolition) and site restrictions that add complexity and increase fatal accidents. Also included were the design complexity as for example the prioritization of aesthetic aspects and the level of construction, which increase fatal accidents like falls from height (these accidents represent 50% of fatal accidents from 1996-1997 to 2007-2008 according to HSE (2009)). Fang, Huang, et al. (2004) included in project nature the following variables: project size, complexity of construction and project management, trying to find relationship with SP.

Taking into account all previous works, we propose to include in site complexity factor the variables showed of complexity of the project, size of site and stage characteristics.

F2) Organisational structure resources

In our research we understand this factor as promoters and constructor’s resources depending mainly on the type of firm, specifically their role and site management resources.

The literature has addressed the relationship among a variety of organization structure elements and results on H&S. One of the organization variables receiving more attention has been company size. For example Liao & Perng (2008) found that more size is related with more probability of accidents. This evidence has been refuted by a large number of studies where the opposite was found (Camino López, Ritzel, Fontaneda González, & González Alcántara, 2011; Camino López, Ritzel, Fontaneda, & González Alcantara, 2008; Holte, Kjestveit, & Lipscomb, 2015; Pérez-Alonso, Carreño-Ortega, Callejón-Ferre, & Vázquez-Cabrera, 2011). Specifically, there seems to be a lot of evidence showing that small companies’ size and private projects with low budgets, have a strong correlation with accidents (Cheng, Leu, Lin, & Fan, 2010)

Fang, Huang, et al. (2004) linked organisational structure as having a significant relationship with H&S on construction sites. In their organisational factor they included safety supervisors, their authority, involvement of contractor management, foreman authority and size of the crew.

Others researches have linked ownership with construction safety (Ros et al., 2013; Baxendale & Jones, 2000; Behm, 2005; Hinze et al., 2013; Xinyu Huang & Hinze, 2006b, 2006b) and defined several practices (mainly related with the hiring and with the promotion technical controls on sites) associated with better H&S. Wu et al.'s (2015) research, about prospective SP evaluations on sites, linked different types of construction companies according to scale, ownership and business strategy, with different levels of SP.

In line with all these empirical studies, we propose to include in our organisational structure resources factor two different variables reflecting promoter’s and constructor’s resources respectively.

F3) Complexity of organisational design

Subcontracting and the coexistence of different agents, each of them with their own characteristics and role, add complexity and make the site more difficult to manage. In this sense, the researchers have found that more levels of subcontracting are associated with low levels of safety (Hinze et al., 2013; López-Alonso, Ibarrondo-Dávila, Rubio-Gámez, & Munoz, 2013; Manu, Ankrah, Proverbs, & Suresh, 2013; Swuste et al., 2012; Yung, 2009). In another research López-Alonso et al. (2013) have found evidence that a direct connection exists between the average number of accidents and the total number of workers, the average number of subcontractors and the health and safety budget. Fang, Huang, et al. (2004) also included the quantity of workers on site and the number of subcontractors within the project nature factor considered in their research. However, they did not find this factor significantly related to SP. Manu et al. (2010) included in their CPF some references to the complexity of contractual system and players (different agents involved in procurement), and the authors identified that this was responsible of adding risk in the construction sector. They also found that subcontracting was linked with accidents and so do the influence of multiple agents.

Based on these studies, we propose that our complexity of organization design factor included a variable to capture the internal organizational structure and other variable to cover different aspect of the job planning and design.

F4) Safety management resources

We understand this factor as promoter resources, preventive functions of structure and health and safety plan (H&SP).

The preventive functions of the persons in charge are a problem especially for small companies because the limitations to get specialised human resources who deal only with H&S problems. Obligations in the field of H&S in small and medium-sized enterprises are usually delegated to one employee in addition to other responsibilities (Jarvis & Tint, 2009). This is a problem that directly affects the possibility of control H&S. Baxendale & Jones (2000) arrived to a similar conclusion as they affirm that the general answer must be to integrate H&S into the day to day activities of the company. Notwithstanding, both studies pointed out the same difficult problem of finding a specialised worker, if she exists and is available, and how to assign her prevention functions. Manu et al. (2013) linked subcontracting with low levels of safety and identified the resources applied by five of the biggest contractors in U.K. One of those contractors insisted on having a non-working subcontractor’s foremen who have direct responsibility for the safety of workers in their trade. Another authors found that safety, injury and illness rates had a clear statistical correlation with the existence of specialised human resources but also considering the clear commitment of management (Abudayyeh et al., 2006). They concluded that supervisors and managers of construction companies can impact the safety on site in a variety of ways.

Turning the attention to the assessment of risk and the management of risk, Mahmoudi et al. (2014) concluded that they are the two most important H&S issues at project level. We can situate in this same line the research of Borys (2012) that founds a connection between previsions of processes and H&S conditions. In the same way, the research of Adam et al. (2009) pointed out that is necessary to make previsions on resources and processes needed to get adequate safety levels. All these studies are in fact considering the most important elements that any H&S plan should incorporate. As it is well known, the H&S plan is the essential reference regarding the obligations on H&S on sites and it is linked with the project.

Safety coordinator, which was even ruled in the European Directive 92/57 CEE, is another factor that is relevant in safety management (Ros et al., 2013).

Fang, Huang, et al. (2004) included the management resource factors composed, among others, of safety inspections (including those carried out by owner’s initiative), safety plan and records. Those safety inspections by owners implicitly are referring to the safety coordination activities which are rule in European legislation. These authors also considered some other items related with involvement of contractor’s manager with site safety, authority of safety supervisor and foreman. All these are related with our factor of safety resources, specifically with the preventive functions of the organisational structure.

To conclude with this factor, according to the evidence reviewed we propose the safety management resources factor to be composed by three different variables: coordination resources, preventive resources and finally, health and safety plan.


We propose that risk level on construction sites is explained basically by the four mentioned organizational factors “Site complexity” (positive relationship), “Organizational structure resources” (negative relationship), “Organization design complexity” (positive relationship) and “Safety management resources” (negative relationship). In other words, those factors can be taken as direct predictors of Site Risk. Therefore, the underlying hypotheses of our structural model can be stated as follow:

H1: Site complexity (e.g. bigger sites, more height, etc.) increase risk on site.

H2: Organizational structure resources (e.g. promoter and constructor’s resources) decrease risk on site.

H3: Organization design complexity (e.g. bigger number of workers, more number of firms, more levels of subcontracting, etc.) increase risk on site.

H4: Safety management resources (e.g. coordinator, preventive functions and H&SP) decrease risk on site.


Data collection

All our data come from specific field work and they were collected through direct observation on site with own toll. This tool lets us obtain and analyse relevant and direct information of building construction sites. With this tool we obtain information such as site characterisation, promoter and contractor characteristics, available documentation, general conditions, the specific risks on the stage, the protections, the processes, the auxiliary resources and the machinery.

The toll registers responses and assessments to a total of 82 questions or items. 60 of them are the source to build our 10 risk variables and the remaining 22 items to compose our 10 organizational variables. In next section we explain in more detail how we operationalise all the variables in our model.

Organizational variables

Based on literature review we have built a total of 10 organizational variables that we also confirmed with an expert panel consultation. These organizational variables were classified as pertaining to one of four general factors or latent variables. Each organizational variable has been composed with some relevant items driven from empirical studies from a pool of a total of 22 items. Therefore, the general structure of our independent organizational variables is formed by 4 latent variables or factors with their associated 10 observed variables which in turn have been built upon their corresponding item from a set of 22 items (see Table in Appendix B for a detail of the items and measurement scales). The first factor “F1 Site complexity” is measured with two variables and six items: complexity of project (OV1) (based on three items describing whether the construction is new or a reform, building configuration as typology, special environmental conditions); size of the site (OV2) (based on one item reflecting number of floors); Stage characteristics (OV3) (based on two items that reflects the complexity of main and secondary work stage). The second factor “F2 Organisational structure resource” is composed by two variables and four items: promoter’s resources (OV4) (based in one item capturing type of promoter firm in terms of their amount of resources); constructor’s resources (OV5) (based on three items type of constructor firms according to their resources, resources depending on constructor’ role -subcontractor, contractor or promoter-constructor-, and site management structure that reflects who is in charge on site). The factor “F3 Organisational design complexity” is measured through three variables and seven items: internal organisational structure (OV6) (based on three items type of contracting -unique contractor or several contractors-, number of companies and level of subcontracting); job planning and design (OV7) (based on four items number of works, employee location assignment –where is the worker-, total number of workers and finally the ratio between own workers and total number of workers). Finally, the last factor “F4 Safety management resources” is composed by three variables and five items: coordination resources (OV8) (composed by two items describing the designation and documented work of coordinator); preventive functions (OV9) (composed by one item assessing the preventive involvement and functions of human resources); H&S plan (OV10) (composed by two items analysing the presence and appropriateness of H&S plan previsions).

Risk variables

As we have already mentioned, our goal is to analyse the relationships among organizational factors and risk level in construction sites. Therefore in our theoretical model our dependent variable is a construction site risk index (SRI) as a proxy of risk level. We propose to calculate the SRI as the average of ten different risk variables, based on both empirical literature review on risk assessment at construction sites and on authors’ technical judgement which is grounded with more than 25 years of professional and technical experience in H&S inspections. Those risk variables are derived directly from the work conditions at sites. They are composed from a group of items which are shown in Table 6 which are the main source of data to measure our SRI index.

The observed values in all risk items were collected using our own toll described before. According to our technical criteria and in order to have a more accurate assessment of risk in construction sites, we consider very important to differentiate the two more relevant ambits at sites, the general conditions affecting the whole construction and the specific condition of the construction stage. This is because the physical environment of sites is a complex mix of different elements that we need to assess independently in terms of risk quantification. Most of the available tools to assess the risk on sites do not consider this differentiation. In other words, we can disaggregate the information about the risk distinguishing among general risks and specific ones, which in turn allow us to propose more focused interventions. Another issue considered as relevant to measure risks is to obtain specific and explicit information about the existing processes and protections on sites. Those issues are specifically mentioned in the theoretical literature (Hollnagel, 2008; Swuste et al., 2012)but not explicitly considered in empirical risk assessment tools because they only measure each risk level without giving specific information of the protection and processes behind the event under assessment. Having that information disaggregated let us to propose direct interventions to correct the risks. Finally, we believe necessary to have specific information as to whether there is evidence about the accomplishment of the H&S plan, which is the most important reference of health and safety previsions that should be achieved on site. This is not included in any of the available risk assessment tools and we propose to include it explicitly.

Following we explain briefly the intuition of each variable:

RV1. Health and safety plan. This indicator measures accomplishment of H&SP, this document is the legal principal reference of H&S on site.

RV2. General condition of the construction work: This variable measures the general conditions of the site, without looking at the specific characteristics of the stage. It is formed from enclosures, circulations (including tidiness, cleanliness and illumination), the signalling and the electrical system.

RV3. General conditions of the collective protections: This variable measures the general collective protections of site, without considering the current stage.

RV4. Specific conditions of phase access; This variable measures the specific conditions of access to the main stage site, independent from general access.

RV5. Fall from height: This variable measures the risk of falling from height in the observed site.

RV6. Other risks: This variable identifies the existence of other risk on the stage, including access and its incidence with the risk of falling from a height.

RV7. Process: This variable tries to identify whether or not the sequence of forecasted interventions is adequate and whether or not they are performed in the correct way, following the forecast process.

RV8. Collective protections: This variable assesses the collective protections in the development stage.

RV9. Individual protections: This variable measures the individual protection for falling from height at development stage.

RV10. Auxiliary resources and machinery. This variable evaluates the adequacy and assessment of the installation of different resources and machinery at the construction site.

The risk variables RV1, RV2, RV3, RV4, RV5 and RV7 were calculated as the average of their items that were all measured with the four level scale. The remaining risk variables, RV6, RV8, RV9 and RV10, combined items that were measured with both type of scales. The calculation of their value followed a stringent criteria for aggregation that penalized for higher risk, making the variable value equal to the value of that item with the highest risk.

Having the observed values of those risk variables for all sites in our sample, we have built the SRI. SRI is calculated as the arithmetic mean of the 10 risk variables and for this reason it has the same four level scale signalling the risk group class. The summary statistic all risk variables and SRI are in Appendix H.


A recent study in the field H&S in construction sector has proposed to use SEM methodology to test several hypotheses regarding the drivers of safety performance (Wu et al., 2015). SEM is used to represent interactions and relationships between a set of observed variables and other variables not observed or latent variables. In a similar way, we propose to test our set of hypotheses using this methodology. Our goal is to test our proposed model and hypotheses (see complete model specification in Appendix A). As it is known, the results of SEM gives evidence about the goodness of fit of the data to the theoretical model proposed. SEM also estimates the strength and significance of the relationships among the variables of the model.

After having determined descriptive statistics and correlations between observed variables, we have performed, multivariate normality tests to assess the underlying statistical assumptions of SEM estimation methods using PRELIS 2 program. Although our data did not manage to fulfil the assumption of multivariate normality, a small degree of deviation (skewness and kurtosis z values below |1.00|) did not invalidate the use of the maximum likelihood method with LISREL 8.80 software (Jöreskog, K.G. & Sörbom, D., 2006). Covariance errors between items were not implemented for the estimated model.

To assess overall fit of the model, c2, the relative/normed c2 to degrees of freedom (df) ratio, the Root Mean Square Error of Approximation (RMSEA) and its 90% Confidence Interval (with a p-value related to RMSEA<.05), the Standardized Root Mean Squared Residual (SRMR), the Comparative Fit Index (CFI), the Normed Fit Index (NFI) and the Goodness of Fit Index (GFI) were the used indices. A model can be considered to fit the data if c2 is non-significant, c2/df <3, RMSEA<.05, SRMR<.08, and CFI, NFI and GFI≥.95 (Hu & Bentler, 1999; Schreiber, Stage, King, Nora, & Barlow, 2006). Finally, to test single parameters, the 5% significance criterion was adopted (i.e., t-value of parameters of 2.00).

Structural model

As in any SEM analysis we have two different models: the measurement model and the structural model. As preliminary results showed that the observed variable OV3 (stage characteristics) was not significant we removed this variable from the analysis and fitted the model in figure 1.

*p<.05 **p<.01

Figure 1. SEM model on SRI, standardized path coefficients, significance levels and goodness of fit indices.

Results for our model (Figure.1) showed an overall adequate fit to data (c2=93.16, df =26, p<.001), c2/df was lower than 5 and near than 3 (3.58), RMSEA stood slightly above at the cutoff of 0,5 for good fit (RMSEA=.058) with a p(RMSEA<.05)=.13 (CI90% RMSEA: .045; .07), both CFI (.98) and GFI (.98) were also indicative of good fit as SRMR (.036). Thus, except for the case of RMSEA all the other indices met the recommended limits to consider our model as good fitted. However, those RMSEA values below 0,8 can be considered as a not bad fit (Wu et al., 2015). We conclude that our model can be considered as acceptable. All path coefficients were statistically significant (p<.01). As expected, site complexity had a positive direct effect on SRI (.55). Organisational structure resources had a very strong and negative direct effect on SRI (-.95). We found also a strong positive direct relationship between organisational design complexity and SRI (.82). Finally, safety management resources were directly and negatively related to SRI (.55). In sum, our four hypotheses stated above (H1-H4) were confirmed. Looking into our results it may be concluded that our model of organisational latent factors explained 20.00% of the variance of SRI.

Discussion of results

The aim of the present research was to test the relations between organizational complexity and resources represented by four latent variables: site complexity, organization structure resources, organization design complexity and safety management resources as potential predictors for risk level on construction sites.

Site complexity is a latent variable composed by complexity of project (OV1) and size of site (OV2). According to our results, site complexity had a positive direct effect on RSI. This result is consistent with the evidence found in previous researches (Fang, Huang, et al., 2004; Forman, 2013; Hatipkarasulu, 2010; Hon et al., 2010; Manu et al., 2010). At the measurement model of our SEM, the latent variable of site complexity explained 29.29% of the variance of complexity of the project and 44.89% of the variance of size of site, which are the two observed variables that composed it. These two last variables represented the complexity measured through type of works, configuration and environment conditions (items of OV1), and size assessed by number of floors (OV2).

As we have reported in the previous section, the strongest effect on RSI is caused by the latent variable named organizational structure resources. From our fitted model follows that this latent variable explains the 46.24% of the variance of promoter resources (OV4) which is based on only one item reflecting type of promoter’s resources and the 23.04% of the variance of contractor’s resources (OV5) conformed with the items of contractor’ type, role and site management structure. On one hand, the implied relationship between promoter (OV4) and SRI in our model is consistent with previous research where it has been found that type of promoter has relevant influence on H&S on site (Baxendale & Jones, 2000; Behm, 2005; Hinze et al., 2013; Ros et al., 2013; Xinyu Huang & Hinze, 2006b, 2006b). On the other hand, the specific relationship between contractor’s resources (OV5) and RSI is more controversial since previous research has not found concluding evidence although there is empirical evidence that relates organizational structure with H&S in general terms (Camino López et al., 2008; C.-W. Cheng et al., 2010; Holte et al., 2015; Pérez-Alonso et al., 2011). Wu et al. (2015, p.71) obtained similar conclusions comparing different levels of safety performance with different types of constructions companies in China, specifically they found the best level of safety performance in Sino-foreing joint ventures, followed by state owned enterprises and finally private enterprises with the lowest level of safety.

We also found a strong effect of latent variable organizational design and complexity on SRI (path coefficient of .082). This latent variable is formed by observed OV6 and OV7. The internal organization of the different companies on site (OV6) is composed by the type of contracting (one o more contractors), number of companies and level of subcontracting. Results imply that the complexity of organizational design explains the 46.24% of the variance of OV6. Previous research have found that more subcontracting leads to worst levels of safety (Hinze et al., 2013; López-Alonso et al., 2013; Manu et al., 2013; Swuste et al., 2012; Yung, 2009), our results are consistent with this evidence. There is no previous research to compare our results regarding the effects on risk either through the number of contractors or the number of companies.

The other observed variable OV7 (job planning and design) that loads on the latent factor of organizational design complexity, contained the items of number of works, location of workers, total number of workers and ratio of own constructor’s worker over the total number of workers. According to our results, the unobserved factor of complexity of organizational design only explained the 12.25% of the variance of OV7. Among the items that compose OV7, only total number of workers on site was analysed in previous researches, where has been found a negative effect on H&S on sites, which is consistent with our results (Fang, Huang, et al., 2004;López-Alonso et al., 2013;).

Finally, the latent variable related with safety management resources is composed by coordination resources (i.e. designation of a H&S coordinator and his documented work) (OV8), preventive functions of the structure (OV9), and presence and appropriateness of the H&S plan (OV10). Our results yielded that this latent variable explains the 32.49% of variance of OV8, the 54.76% of variance of OV9 and the 21.16% of variance of OV10. Our results are consistent with previous research referring to the preventive functions of the structure (Baxendale & Jones, 2000; Borys, 2012; Fang, Huang, et al., 2004; Jarvis & Tint, 2009; Manu et al., 2013). There is not previous research relating documented work of H&S coordinator and the appropriateness of H&S plan with levels of safety conditions on site. Despite preventive functions of the structure has the highest path coefficient in the relationship with safety management resources (.74), H&S coordinator and H&S plan have also a relevant path coefficients in this relationship, .57 and .46 respectively.


Seams construction sector is typically one of the most high risk industries, there is a broad concern in finding approaches to address this problem. In order to control the risk at reasonable levels, one of the most prominent approaches consists in obtaining a set of leading indicators of safety conditions (i.e. predictors) before the accident event occurs (Hinze et al., 2013). Our study adopts this approach to analyse the relationship between four organizational variables, representing complexity and resources, and the risk level on site. Specifically, we have proposed that higher complexity in both the site and the organizational design will have a negative impact on safety levels. We have also proposed that higher amount of resources behind the organizational structure and supporting safety management activities will reduce risk. A structural equation model has been proposed to test these hypotheses and we have fitted the model using data collected during 957 direct inspections on construction sites.

As a result of our research the following general conclusions can be drawn. In first place, we found that our four organizational variables (site complexity, organizational structure resources, organizational design complexity and safety management resources) have direct and relevant impact on an index reflecting level of risk (SRI) as our hypotheses proposed. The relevancy of those findings is that the organizational variables considered can be used as predictor of risk on site and therefore they have important implications on H&S intervention as we propose below.

In second place, our analyses revealed that all our relationships (hypotheses) were statistically significant at the level of 0.01 and with the expected sings. We have obtained that the most important factor to explain the risk on sites is the organizational structure resources (F2 in Figure 2) with a direct and negative path coefficient of -0.95. We have also found that the organizational design complexity (F3) has also an important positive and direct effect on risk, specifically the estimated path coefficient is 0.82. Weaker effects on site risk were found for the factor of site complexity (F1), with a positive path coefficient of 0.55 and safety management resources (F4) having a negative path coefficient of -0.55.

In summary, our results showed that the risk on site is mainly affected by organization structure and organizational design complexity, and the most relevant items variables within those two factors were promoters’ and constructors’ resources (OV4 and OV5) and organizational structure (OV6) respectively. Some important implications can be derived from these findings. On one hand, in order to have more safety in construction sites a reinforcement of the promoter role having more professional agents endowed with more resources is needed because it will increase the involvement of promoters with H&S issues. Regarding constructor’s resources (OV5) our results suggest that improving the constructors’ endowments of resources, reinforcing constructor’s role on leading the effective works on sites and strengthening site management structure might improve safety on site. All those goals can be achieved having more professionalised companies with more stable structures and well trained and informed workforce about H&S issues. It is equally important to assure an active presence and control on works by contractors and the assignment of the appropriate human resources to be explicitly present on site.

On the other hand, since organizational structure variable (OV6) is composed by type of contracting, number of companies and level of subcontracting which affect negatively the risk level on site, might be necessary to consider whether construction normative and regulations should address these questions. Measures such as limiting the level of subcontracting are already implemented, but other interventions which are not currently undertaken might be, for example, limiting the number of contractors or the total number of companies on site. Of course, which are the right limits along with the efficiency of such measures should be evaluated both in economic and social terms.

Following with our estimated model, we observed that safety management resources had an impact on risk on site. This factor was measured upon coordination resources (OV8), preventive functions (OV9) and health and safety plan (OV10). The most remarkable effect found is for preventive functions (OV9) in comparison to the effects of the other two variables. We conclude that is essential to assume preventive functions. Connecting with results obtained for factor F2 (organizational structure resource), seems that in order to have acceptable levels of safety would be not enough simply having the figure of a foreman or manager because the relevant thing is whether or not this foreman or manager is really assuming and developing the right preventive functions. A straightforward implication can be extracted in the sense that performance appraisal of the manager’s H&S functions is as or even more important than adding these functions to the job design of the site manager or foreman.

Although our hypothesized model has obtained and adequate fit, the main theoretical limitation of the present research is that we have addressed only one side of the problem to explain the level of risk, let us say the life conditions on sites, ignoring the side determined by workers behaviours (attitudes, climate, culture, etc.). As we have commented in the literature review, this behavioural aspect of risk has been proposed and empirically analysed in other studies. We think that future models intended to explain risk and to propose safety interventions should consider jointly both faces of the same coin, the behavioural and the “material” (i.e. life conditions) nature of risk.

Other limitation of our study is that the majority of the sites in our sample are small and medium local constructions firms with low levels of resources. Of course, this supposes a limitation to generalise our results to the whole sector or other geographical regions.

Despite the limitations we still believe this research has relevant contributions. In first place, we have used field data from a large number of sites collected directly from the work scenarios while we were developing safety inspections. In second place, we have found evidence supporting that organizational factors, as complexity and resources, are good predictors of level of risk. This is important to develop tools aimed to obtain indicators that provide risk information in advance to risk exposure or to the accident occurrence (this kind of indicators are commonly known as leading indicators in contrast to lagging indicators which are based on past accidents). Finally, having in mind our results we can propose interventions and design safety campaigns to direct the adequate actions towards those organizational dimensions which might be higher sources of risk.

Other considerations

The present paper is a resume from complete one, this is the explanation of the misses of cited appendixes.

Referencias bibliográficas

Abudayyeh, O., Fredericks, T. K., Butt, S. E., & Shaar, A. (2006). An investigation of management’s commitment to construction safety. International Journal of Project Management, 24(2), 167–174. http://doi.org/10.1016/j.ijproman.2005.07.005

Adam, J. M., Pallarés, F. J., & Calderón, P. A. (2009). Falls from height during the floor slab formwork of buildings: Current situation in Spain. Journal of Safety Research, 40(4), 293–299. http://doi.org/10.1016/j.jsr.2009.07.003

Ai Lin Teo, E., & Yean Yng Ling, F. (2006). Developing a model to measure the effectiveness of safety management systems of construction sites. Building and Environment, 41(11), 1584–1592. http://doi.org/10.1016/j.buildenv.2005.06.005

Antonio, R. S., Isabel, O.-M., Gabriel, P. S. J., & Angel, U. C. (2013). A proposal for improving safety in construction projects by strengthening coordinators’ competencies in health and safety issues. Safety Science, 54, 92–103. http://doi.org/10.1016/j.ssci.2012.12.004

Baxendale, T., & Jones, O. (2000). Construction design and management safety regulations in practice--progress on implementation. International Journal of Project Management, 18(1), 33–40. http://doi.org/10.1016/S0263-7863(98)00066-0

Behm, M. (2005). Linking construction fatalities to the design for construction safety concept. Safety Science, 43(8), 589–611. http://doi.org/10.1016/j.ssci.2005.04.002

Borys, D. (2012). The role of safe work method statements in the Australian construction industry. Safety Science, 50(2), 210–220. http://doi.org/10.1016/j.ssci.2011.08.010

Camino López, M. A., Ritzel, D. O., Fontaneda González, I., & González Alcántara, O. J. (2011). Occupational accidents with ladders in Spain: Risk factors. Journal of Safety Research, 42(5), 391–398. http://doi.org/10.1016/j.jsr.2011.08.003

Camino López, M. A., Ritzel, D. O., Fontaneda, I., & González Alcantara, O. J. (2008). Construction industry accidents in Spain. Journal of Safety Research, 39(5), 497–507. http://doi.org/10.1016/j.jsr.2008.07.006

Cheng, C.-W., Leu, S.-S., Lin, C.-C., & Fan, C. (2010). Characteristic analysis of occupational accidents at small construction enterprises. Safety Science, 48(6), 698–707. http://doi.org/10.1016/j.ssci.2010.02.001

Cheng, E. W. L., Ryan, N., & Kelly, S. (2012). Exploring the perceived influence of safety management practices on project performance in the construction industry. Safety Science, 50(2), 363–369. http://doi.org/10.1016/j.ssci.2011.09.016

Donaghy, R. (2009). One-death-is-too-many. Secretary of State Report UK.

Fang, D. P., Huang, X. Y., & Hinze, J. (2004). Benchmarking Studies on Construction Safety Management in China. Journal of Construction Engineering & Management, 130(3), 424–432. http://doi.org/10.1061/(ASCE)0733-9364(2004)130:3(424)

Fang, D. P., Xie, F., Huang, X. Y., & Li, H. (2004). Factor analysis-based studies on construction workplace safety management in China. International Journal of Project Management, 22(1), 43–49. http://doi.org/10.1016/S0263-7863(02)00115-1

Fernández-Muñiz, B., Montes-Peón, J. M., & Vázquez-Ordás, C. J. (2009). Relation between occupational safety management and firm performance. Safety Science, 47(7), 980–991. http://doi.org/10.1016/j.ssci.2008.10.022

Forman, M. (2013). Inertia and change: lean construction and health and safety work on construction sites. Construction Management & Economics, 31(6), 647–660. http://doi.org/10.1080/01446193.2013.765953

Ghasemi, F., Mohammadfam, I., Soltanian, A., Mahmoudi, S., & Zarei, E. (n.d.). Surprising Incentive: An Instrument for Promoting Safety Performance of Construction Employees. Safety and Health at Work. http://doi.org/10.1016/j.shaw.2015.02.006

Grabowski, M., Ayyalasomayajula, P., Merrick, J., & McCafferty, D. (2007). Accident precursors and safety nets: Leading indicators of tanker operations safety. Maritime Policy and Management, 34(5), 405–425. http://doi.org/10.1080/03088830701585084

Hatipkarasulu, Y. (2010). Project level analysis of special trade contractor fatalities using accident investigation reports. Journal of Safety Research, 41(5), 451–457. http://doi.org/10.1016/j.jsr.2010.08.005

Hinze, J., Thurman, S., & Wehle, A. (2013). Leading indicators of construction safety performance. Safety Science, 51(1), 23–28. http://doi.org/10.1016/j.ssci.2012.05.016

Hoewijk, V. R. (1988). De betekenis van de organisatiecultuur: een literatuuroverzicht (The meaning of organisational culture: an overview of the literature). M & O, Tijdschrift Voor Organisatiekunde En Sociaal Beleid 1, 4– 46.

Hollnagel, E. (2008). Risk + barriers = safety? Safety Science, 46(2), 221–229. http://doi.org/10.1016/j.ssci.2007.06.028

Holte, K. A., Kjestveit, K., & Lipscomb, H. J. (2015). Company size and differences in injury prevalence among apprentices in building and construction in Norway. Safety Science, 71, Part C, 205–212. http://doi.org/10.1016/j.ssci.2014.01.007

Hon, C. K. H., Chan, A. P. C., & Wong, F. K. W. (2010). An analysis for the causes of accidents of repair, maintenance, alteration and addition works in Hong Kong. Safety Science, 48(7), 894–901. http://doi.org/10.1016/j.ssci.2010.03.013

HSE. (2009). Health and Safety Executive. Underlying Causes of Construction Fatal Accidents – A Comprehensive Review of Recent Work to Consolidate and Summarize Existing Knowledge, Phase 1 Report. Construction Division. Her Majesty’s Stationary office, Norwich.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. http://doi.org/10.1080/10705519909540118

Jarvis, M., & Tint, P. (2009). The Formation of a Good Safety Culture at Enterprise. Journal of Business Economics and Management, 10(2), 169–180.

Jöreskog, K.G., & Sörbom, D. (2006). LISREL 8.80 for Windows [Computer Software]. Lincolnwood, IL: Scientific Software International, Inc.

Khanzode, V. V., Maiti, J., & Ray, P. K. (2012). Occupational injury and accident research: A comprehensive review. Safety Science, 50(5), 1355–1367. http://doi.org/10.1016/j.ssci.2011.12.015

Liao, C.-W., & Perng, Y.-H. (2008). Data mining for occupational injuries in the Taiwan construction industry. Safety Science, 46(7), 1091–1102. http://doi.org/10.1016/j.ssci.2007.04.007

López-Alonso, M., Ibarrondo-Dávila, M. P., Rubio-Gámez, M. C., & Munoz, T. G. (2013). The impact of health and safety investment on construction company costs. Safety Science, 60, 151–159. http://doi.org/10.1016/j.ssci.2013.06.013

Mahmoudi, S., Ghasemi, F., Mohammadfam, I., & Soleimani, E. (2014). Framework for Continuous Assessment and Improvement of Occupational Health and Safety Issues in Construction Companies. Safety and Health at Work, 5(3), 125–130. http://doi.org/10.1016/j.shaw.2014.05.005

Manu, P., Ankrah, N., Proverbs, D., & Suresh, S. (2010). An approach for determining the extent of contribution of construction project features to accident causation. Safety Science, 48(6), 687–692. http://doi.org/10.1016/j.ssci.2010.03.001

Manu, P., Ankrah, N., Proverbs, D., & Suresh, S. (2013). Mitigating the health and safety influence of subcontracting in construction: The approach of main contractors. International Journal of Project Management, 31(7), 1017–1026. http://doi.org/10.1016/j.ijproman.2012.11.011

Mohamed, S. (1999). Empirical investigation of construction safety management activities and performance in Australia. Safety Science, 33(3), 129–142. http://doi.org/10.1016/S0925-7535(99)00028-4

Pérez-Alonso, J., Carreño-Ortega, Á., Callejón-Ferre, Á. J., & Vázquez-Cabrera, F. J. (2011). Preventive activity in the greenhouse-construction industry of south-eastern Spain. Safety Science, 49(2), 345–354. http://doi.org/10.1016/j.ssci.2010.09.013

Rozenfeld, O., Sacks, R., Rosenfeld, Y., & Baum, H. (2010). Construction Job Safety Analysis. Safety Science, 48(4), 491–498. http://doi.org/10.1016/j.ssci.2009.12.017

Schreiber, J. B., Stage, F. K., King, J., Nora, A., & Barlow, E. A. (2006). Reporting Structural Equation Modeling and Confirmatory Factor Analysis Results: A Review. The Journal of Educational Research, 99(6), 323–337. http://doi.org/10.3200/JOER.99.6.323-338

Sgourou, E., Katsakiori, P., Goutsos, S., & Manatakis, E. (2010). Assessment of selected safety performance evaluation methods in regards to their conceptual, methodological and practical characteristics. Safety Science, 48(8), 1019–1025. http://doi.org/10.1016/j.ssci.2009.11.001

Sparer, E. H., & Dennerlein, J. T. (2013). Determining safety inspection thresholds for employee incentives programs on construction sites. Safety Science, 51(1), 77–84. http://doi.org/10.1016/j.ssci.2012.06.009

Swuste, P., Frijters, A., & Guldenmund, F. (2012). Is it possible to influence safety in the building sector?: A literature review extending from 1980 until the present. Safety Science. http://doi.org/10.1016/j.ssci.2011.12.036

Toellner, J. (2001). Improving Safety & Health Performance: Identifying & Measuring Leading Indicators. Professional Safety, 46(9), 42–47.

Törner, M., & Pousette, A. (2009). Safety in construction - a comprehensive description of the characteristics of high safety standards in construction work, from the combined perspective of supervisors and experienced workers. Journal of Safety Research, 40(6), 399–409. http://doi.org/10.1016/j.jsr.2009.09.005

Wilson, H. (1989). Organisational behaviour and safety management in the construction industry. Construction Management and Economics, 7, 303–319.

Wu, X., Liu, Q., Zhang, L., Skibniewski, M. J., & Wang, Y. (2015). Prospective safety performance evaluation on construction sites. Accident Analysis & Prevention, 78, 58–72. http://doi.org/10.1016/j.aap.2015.02.003

Xinyu Huang, & Hinze, J. (2006). Owner’s Role in Construction Safety. Journal of Construction Engineering & Management, 132(2), 164–173. http://doi.org/10.1061/(ASCE)0733-9364(2006)132:2(164)

Yung, P. (2009). Institutional arrangements and construction safety in China: an empirical examination. Construction Management & Economics, 27(5), 439–450. http://doi.org/10.1080/01446190902855633

Papers relacionados

ORP 2015
Francisco José
Forteza Oliver
Universidad Illes Balears
ORP 2015
ORP 2015
Ojados González
Universidad Politécnica de Cartagena
ORP 2015
Instituto Politécnico Nacional y DIE CINVESTAV