Safety Climate Analysis at Power Station
Basel Marhoon - University of Bahrain 2016
Safety is the main important issue for all worker in the industrial sector. Safety contain many subjects, safety culture and safety climate are one of those subjects and they are concepts that today considered as one of the main concern across the industrial sector. As simple definition, the safety culture is “Making sure people are not harmed is how we do things here”.
This work evaluate the safety climate as it is reflection for the safety culture and easier to evaluate. The evaluation done for private power companies located in Bahrain. First plant (plant A) is for power generation, second plant (plant B) is for power generation and water production. Both plants are belong to the same mother company. The evaluation done through questionnaire including all the staff (around 200). The questionnaire designed by the help of the health and safety department of the mother company. Also, by interviews that used to support the questionnaire. Analysis has been done by different statistical methods using Microsoft excel software.
The results of safety climate evaluation showed that the overall level of safety for is above average (3.72 of 5). In addition, the safety level for plant A (3.93) is higher than plant B (3.49). By studying the safety level for different departments, four groups were identified, which are: 1) Plant A daytime workers, 2) Plant A shift workers, 3) Plant B daytime workers, and 4) Plant B shift workers. The level of safety for daytime workers were higher than shift workers. Also, the study reveal some factors that affect the safety climate level, such as, the nature of the work, the experience and the number of attended workshops.

1. Introduction
Part of human nature is the need to safety, so the issue of safety is a human concern by nature. In the industrial sector, where there will be significant hazards occurring, the need for safety requires the utmost attention. It is important to feel safe while working and to have a safety system in place, which has reflective indicator in order to improve it. It is a crucial matter to everyone.
In the recent years there has been movement away from safety measures based on retrospective data which called ‘lagging indicators’ such as fatalities and time lost in accidents, towards what is called ‘leading indicators’ such as safety audits or measurements of safety climate. The leading indicators can enhance the process of weakness finding without risking the failure of the system (Flin, Mearns, O'Connor & Bryden, 2000).
So studying the safety culture can be strong method to detect the weaknesses in the safety system in order to prevent any occurrence of unsafe events in the future.
Identifying the level of safety culture is difficult and takes a long time, so more common is to evaluate the safety climate which is considered as reflection or snap shot of the safety culture so safety climate can be considered as an alternative safety indicator, (Schein, 2010; Guldenmund, 2010).
The improvement process for the safety culture include: a) determining the current level of safety climate b) deciding the wanted level of safety climate c) creating action plan in order to improve the safety climate level.
This study included a safety climate survey in power generation and water production industry in Bahrain, involving two plants, plant A for power generation, and plant B for power generation and water production. Both plants belong to same mother company.
The safety climate was studied using safety climate questionnaire and interviews.
The aim of this study is to 1- To evaluate the safety climate level. 2- To find out the factors effecting the safety level. 3- To detect sub-climate existence between departments. 4- To find out the weaknesses in the safety system. 5- To decide on the wanted target for the safety climate level according to the management. 6- To give recommendations for improving the existing safety level. 7- To derive new definition for safety culture based on the employees understanding.

2. Method
The general steps of this work starts with performing a literature review to get an idea about the targeted company, safety culture, safety climate and the methods to evaluate safety climate. Next a meeting arranged with the general manager and head of health, safety and environment (HSE) department. Then, the measuring tool design process has been done through multi stages. The final step was conducting questionnaire to evaluate the safety climate level and performing interviews to get better understanding of the questionnaire results. 3.1. Questionnaire
The main measuring tool to evaluate the safety climate is the questionnaire, (Cooper, 2000; Guldenmund, 2010; CANSO, 2008). So prior to starting the evaluation, a meeting had been done with the company safety department representatives and managers of other departments. The meeting done to compare different questionnaires. One questionnaire had been chosen as the main structure to formulate a questionnaire fit to the power plant.
Then many modifications has been done by comparison with other questionnaires and by seeking feedback of HSE department.
The validation process had been done through two steps. First step, seeking feedback of HSE department as expert in the field. Second step, by performing pilot study to improve the questionnaire.
A pilot study had been conducted in order to identify any mysterious questions, spot the weaknesses, insure easy wording and estimate the time required for questionnaire completion. The pilot study included 20 participant. According to Connelly (2008), 10% of the total population is ideal to be used for pilot study. The pilot study repeated two times; every time, notes were taken to modify the questionnaire, then the final version had been achieved. The average time required to finish the questionnaire were calculated. The time was 10.6 minutes.
In this work, to check the questionnaire stability, test-retest has been done. The number of participants were taken similar to pilot test and chosen randomly, which is 20 participants. For the time between the two trials, there were no evidence for the correct time (Parsian & Dunning, 2009), so it’s better to not make it long time or very short time. In this work, the time has been set to ten days. The Interclass Correlation Coefficient (ICC) has been used in order to determine the stability of the questionnaire which reflect its reliability. In addition to the stability test, Cronbach alpha coefficient had been calculated also to measure the internal consistency which also reflect the reliability of the questionnaire.
The questionnaire consists of seven dimension and fifty items, all items are five points Likert scale and it consists of positive and negative questions. The positive questions scored from 1 to 5, with 1 for strongly disagree and 5 for strongly agree. The negative questions scored inversely from 5 to 1, with 5 for strongly disagree and 1 for strongly agree. A mean score over 3 is considered positive result since it is the mean value between the highest and lowest scores (NOSACQ-50, 2015; Kink, 2010; Bergh, 2011; Cooper, 2001).
The seven dimensions are: management ‘safety priority and ability’ (9 items), ‘Management safety empowerment’ (7 items), ‘Management safety judgment’ (6 items), ‘Workers’ safety commitment’ (6 items), ‘Workers’ safety priority and risk non-acceptance’ (7 items), ‘peer safety communication, learning, and trust in safety ability’ (8 items) and ‘trust in the efficacy of safety systems’ (7 items).
In addition to the fifty questions, the questionnaire contains background questions to use it in determining the factors that affect the safety climate level.
The sample size around 210 employee which is all the population in order to increase the accuracy of the results. 3.2. Interviews
The secondary tool to evaluate the safety climate are interviews, (CANSO, 2008; Bergh, 2011). Same as the questionnaire, the interview questions were choose by the help of HSE department
The interview questions are based on the questions of the questionnaire. The interview questions were used to seek more explanation to some items in the questionnaire
In this work, the interviews are structured, single, and it consist of open ended questions. It is structured in order to limit the time required for the interview. Also, it is single which increase the level of confidentiality and let the interviewee talk freely. Finally, it consist of open ended questions to seek the opinions of the interviewee. 3.3. The procedure
After designing the questionnaire, it were handed to every employee in both plants personally. Every one received the questionnaire was informed about the aim of the questionnaire and the benefit of this work. The participants had been told to not write their names. Also, they had been told, the material is confidential that nobody can see the individual results. The questionnaire returned back personally as handed.
Around 210 questionnaires were distributed among all the employees. The return rate was 72.3%.
For the interviews, 30 employee were selected using stratified sampling method. Then, they were asked through emails if they like to participate. The rate of responses for the interviews was 33.33% of the chosen employees. The chosen employees were presenting good mixture from the two plants. 3.4. Data analysis
The data collected through the questionnaire were analyzed by Microsoft office Excel software. The raw data were used to calculate the mean value for each dimension for every submission, than the overall and groups mean score were calculated based on the individual results for each dimension (Prasad & Reghunath, 2010, Cooper, 2001). The mean score for all dimensions had been viewed as radar chart.
Every two groups examined for significant difference using T-test with 95% level of confidence and P-value (probability value) less than 0.05 to consider the test is significant.
The internal consistency of each dimension tested using Cronbach alpha coefficient, which called coefficient of reliability, the value of coefficient greater than 0.7 is considered to be ideal and reliable (Pallant, 2007; George & Mallery, 2003, Gliem & Gliem, 2003). Besides using the internal consistency, Interclass Correlation Coefficient (ICC) used to determine the stability of the questionnaire.

3. Results
The overall mean score for both plant is 3.72 on scale of (1 to 5), (Table 1). The mean score for plant A is greater than plant B, where in plant A the mean score is 3.93 and in plant B is 3.49. The lowest mean score obtained in dimension 4 for plant B (2.86).
The result of the T-test analysis shows that there is significant difference in three dimensions, dimension 1 (Management safety priority and ability), dimension 4 (Workers’ safety commitment) and dimension 6 (Peer safety communication learning and trust in safety ability). (Table 2)
Table (1) The mean scores and standard deviations of the mother company | DIM1 | DIM2 | DIM3 | DIM4 | DIM5 | DIM6 | DIM7 | Overall | Mean value | 3.68 | 3.82 | 3.84 | 3.45 | 3.68 | 3.56 | 4.02 | 3.72 | Mean value % | 67% | 70.5% | 71% | 61.25% | 67% | 64% | 75.5% | 68% | Standard Deviation | 0.41 | 0.52 | 0.4 | 0.67 | 0.43 | 0.54 | 0.42 | 0.35 |
DIM=Dimension
Table (2) The variation in the level of safety climate between plant A and plant B Site | n | DIM1 | DIM2 | DIM3 | DIM4 | DIM5 | DIM6 | DIM7 | Plant A | 40% | 3.81 | 3.93 | 3.94 | 4.01 | 3.73 | 3.97 | 4.14 | | | 70.25% | 73.25% | 73.5% | 75.25% | 68.25% | 74.25% | 78.5% | Plant B | 60% | 3.53 | 3.71 | 3.73 | 2.86 | 3.62 | 3.13 | 3.88 | | | 63.25% | 67.75% | 68.25% | 46.5% | 65.5% | 53.25% | 72% | P-value | | 0.035 | NS | NS | 2.14E-10 | NS | 1.57E-8 | NS |
DIM= Dimension, NS= Non-Significant, n= percentage of participants, P-value= result of the T-Test

For the comparison between the workers (3.72) and managers/supervisors (4.13) at both plants, significance difference has been found in three dimensions, dimension 1 (Management safety priority and ability), dimension 2 (Management safety empowerment) and dimension 5 (Workers’ safety priority and risk non-acceptance). (Table 3)
Table (3) The variation in the level of safety climate between the workers and the managers/supervisors at both plants. | n | DIM1 | DIM2 | DIM3 | DIM4 | DIM5 | DIM6 | DIM7 | Overall | Workers | 83% | 3.68 | 3.82 | 3.84 | 3.45 | 3.68 | 3.56 | 4.02 | 3.72 | | | 67% | 70.5% | 71% | 61.25% | 67% | 64% | 75.5% | 68% | Managers/ Supervisors | 17% | 4.32 | 4.27 | 3.92 | 4.13 | 4.26 | 3.88 | 4.16 | 4.13 | | | 83% | 81.75% | 73% | 78.25% | 81.5% | 72% | 79% | 78.25% | P-value | | 0.0002 | 0.048 | NS | NS | 0.002 | NS | NS | |
DIM= Dimension, NS= Non-Significant, n= percentage of participants, P-value= result of the T-Test
By dividing the workers into two main categorize (daytime and shift workers) among the two sites, four groups can be identified. Plant A daytime workers, plant A shift workers, plant B daytime workers, and plant B shift workers. The comparison between shift and daytime showed that daytime workers has higher level of safety climate in both plants. Table 4 showed the results of safety climate analysis for each group.
Table (4) The level of safety climate within plant A and B, for shift workers and daytime workers. Group | DIM1 | DIM2 | DIM3 | DIM4 | DIM5 | DIM6 | DIM7 | Overall | Plant A Day time workers | 4.06 | 4.23 | 4.06 | 4.04 | 3.89 | 3.98 | 4.29 | 4.07 | | 76.5% | 80.75% | 76.5% | 76% | 72.25% | 74.5% | 82.25% | 76.75% | Plant A Shift workers | 3.64 | 3.71 | 3.85 | 3.98 | 3.61 | 3.95 | 4.04 | 3.82 | | 66% | 67.75% | 71.25% | 74.5% | 65.25% | 73.75% | 76% | 70.5% | Plant B Day time workers | 3.84 | 3.91 | 3.81 | 2.83 | 3.75 | 3.39 | 4 | 3.64 | | 71% | 72.75% | 70.25% | 45.75% | 68.75% | 59.75% | 75% | 66% | Plant B Shift workers | 3.28 | 3.54 | 3.67 | 2.89 | 3.51 | 2.92 | 3.79 | 3.37 | | 57% | 63.5% | 66.75% | 47.25% | 62.75% | 48% | 69.75% | 59.25% |
DIM= Dimension

For the factors affecting the safety climate level, four factors had been revealed clearly. 1) As the experience increased, the safety climate level increase; 2) The workers with more workplaces scored higher safety climate level; 3) As the number of safety workshops increased it will result in higher safety climate level; 4) Also the workers with more experience in the plant itself scored higher level of safety climate.

4. Discussion 5.5. Method
The evaluation of internal consistency of the questionnaire’s seven dimensions showed positive results as the alpha coefficient values were between 0.55 and 0.81 with average of 0.73. This result considered accepted, (Pallant, 2007; George & Mallery, 2003, Gliem & Gliem, 2003). Also the stability test showed positive results as it has been found between 0.495 and 0.751 which indicate acceptable stability for the questionnaire, (Kines et al., 2011)
The minimum response rate for a survey to be acceptable, depend on the size of the population, (Basarb, 2011). In this work, 72.3% response rate had been achieved for sample size around 210. According to Basarb (2011), the minimum acceptable response rate for 200 sample size, is 66%. The response rate for the interviews was 33.3% (10 out of 30). 5.6. Results
As overall safety climate level for Al Ezzel o&M, it is considered as above average. The overall mean score of all results is 3.72 on scale of (1 to 5). The mean score for most of the dimensions were higher than 3 which considered positive result, (Prasad and Reghunath, 2010; Bergh, 2011)
By investigating the two sites separately, a significant difference has been found in dimension 1 (Management safety priority and ability), dimension 4 (Workers’ safety commitment) and dimension 6 (Peer safety communication learning and trust in safety ability). That indicate a gap in the safety climate level between the two sites. The answers of the interviews can clarify some reasons behind this gap. The answers state that a difference may exists because the safety enhancement process started in plant A before plant B, alongside with applying of standardized safety rules in plant A before plant B. Other reason, the locating of management in plant A, gave it preference in adapting the safety rules and enhancing the safety communication and reporting. Also, according to Reason (1997), the level of safety climate can be affected by experiencing more accidents. By comparing the findings of Reason 1997 and the finding of this work, the same conclusion can be achieved. Referring to the mother company documents (2015), 65% of the near-misses were occurred in plant B, while plant A has 35%. According to present work results, plant B scored less safety climate level.
By comparing the safety climate between daytime workers and shift workers, daytime workers have more safety climate level than shift workers. The previous studies have showed that working near the production line (in present work, shift workers) will affect the level of safety climate to be lower than the safety climate level of the other workers who work not near to the production (in the present work, daytime workers), (Collins &Gadd, 2002; Milczarek & Najmiec, 2004, Wu et al., 2007; Bergh et al., 2013). This can be seen similar to the present work results. The shift workers are working closer to the production which in this case: gas turbines, steam turbines, energized equipment and fuel gas filled systems etc. Day time workers are working most of the time far of production line, such as workshops and store.
By comparing the safety climate level between the workers and managers, the managers have higher level of safety climate than the workers. One possible reason is that the safety issues discussed between the management and safety department are not communicated effectively to the workers. This analysis is supported by the results of the interviews answers.

5. Conclusion
The results of studying the safety climate of both plants showed that the level of safety for the mother company is above average (3.72 out of 5). The safety level of plant A (3.92) is higher than plant B (3.49). So plant B needs more attention in any safety improvement process.
Two sub climates were detected within the company, daytime workers and shift workers. The sub climate appears obviously more at plant B where the shift workers record less safety climate level.
The weaknesses are few. For the day time workers of plant B, only one dimension showed weakness which is dimension 4 (Workers’ safety commitment). And for the shift workers of plant B, two dimensions showed weaknesses which are dimension 4 (Workers’ safety commitment) and dimension 6 (Peer safety communication learning and trust in safety ability).

6. Recommendation
First, the management should communicate the safety culture definition. It can use the formulated definition in the present work. By communicating the definition, shared understanding to safety culture concept can be created.
The second recommendation is to decide the wanted goal of safety climate level and create a plan for safety culture improvement based on the outcome of the analysis of the questionnaire and the interviews. Also, all employees should be involved in some way in putting the action plan of improvement of safety culture to create shared goal among the employees.
Finally, the improvement has to be followed up by re-evaluating the safety climate level every two or three years using the same questionnaire in order to be able to compare the results.

References
Basarb, D. (2013). Minimum Size for Conducting Training Evaluation Studies. Retrieved from Dave Basarb Consulting: http://www.davebasarab.com/blog/evaluation- survey/minimum-size-for-conducting-training-evaluation-studies/ Bergh, M. (2011). An evaluation of the safety climate at AkzoNobel Site Stenungsund .
Master of Science Thesis, Chalmers University of Technology, Sweden

Bergh, M., Shahriari, M., & Kines, P. (2013). Occupational Safety Climate and Shift
Work. Chemical Engineering Transactions, 31, 403-408.

CANSO Safety Standing Committee (2008). Safety Culture Definition & Enhancement
Process Model. The Civil Air Navigation Services Organization, Netherlands.

Collins, A. M., & Gadd, S. (2002). Safety culture: a review of the literature. Sheffield:
Health and Safety Laboratory, Human Factors Group.

Connelly, L. M. (2008). Pilot studies. Medsurg Nursing, 17(6), 411-2.

Cooper, M. D. (2000). Towards a model of safety culture. Safety Science, 36, 111-136.

Cooper, M. D. (2001). Improving safety culture: A practical guide. London: Hull.

Flin, R., Mearns, K., O'Connor, P. and Bryden, R. (2000). Measuring safety climate: identifying the common features. Safety Science, 34, 177-192.

George, D., & Mallery, P. (2003). SPSS for Windows step by step: A simple guide and reference. 11.0 update (4th Ed.). Boston: Allyn & Bacon.

Gliem, J. A., & Gliem R. R. (2003). Calculating, interpreting, and reporting Cronbach’s
Alpha Reliability Coefficient for Likert-Type Scales. Midwest Research to Practice Conference in Adult, Continuing, and Community Education. The Ohio State University, Columbus, OH.

Guldenmund, F. W. (2010). Understanding and Exploring Safety Culture.Oisterwijk:
Uitgeverij BOXPress.

Kines, P., Lappalainen, J., Mikkelsen, K. L., Olsen, E., Pousette, A., Tharaldsen, J.,
Tómasson, K., & Törner, M. (2011). Nordic Safety Climate Questionnaire
(NOSACQ-50): A new tool for diagnosing occupational safety climate. International Journal of Industrial Ergonomics, 41, 634-646.

Kinik, Y. (2010). Improving Safety Culture: A study performed at KLM Engineering &
Maintenance.

Milczarek, M., & Najmiec, A. (2004). The Relationship between Workers' Safety Culture and Accidents, Near Accidents and Health Problems. International Journal of Occupational Safety and Ergonomics, 25-33.

NOSACQ-50. (2015). Interpreting the Nordic Occupational Safety Climate
Questionnaire NOSACQ-50 resultshttp://www.arbejdsmiljoforskning.dk
/da/publikationer/spoergeskemaer/nosacq-50/how-to-use-nosacq-50/interpreting-nosacq-50-results

Pallant, J. (2007). SPSS Survival Manual. Berkshire: McGraw Hill.

Parsian, N., & Dunning, D. (2009). Developing and Validating a Questionnaire to
Measure Spirituality: A Psychometric Process. Global Journal of Health Science, 1(1), 2-11.

Prasad, S. V., & Reghunath, K. P. (2010). Empirical Analysis of Construction Safety
Climate – A Study. International Journal of Engineering Science and Technology, 2(6), 1699-1707.

Reason, J. (1997). Managing the Risks of Organizational Accidents. Aldershot: Ashgate
Publishing Limited.

Schein, E. H. (2010). Organizational Culture and Leadership. San Francisco: Jossey-
Bass.

Wu, T. C., Liu, C. W., & Lu, M. C. (2007). Safety climate in university and college laboratories: Impact of organizational and individual factors. Journal of Safety Research, 91–102.