1. Introduction

2. Research Significance and Contribution

3. Research Methodology

4. Case Study: Organization Resilience: Lessons Learned from the Oil and Gas Sector in Qatar

4.3. Data Analysis

Data cleaning and preparation is a very important step to ensure that the results are not biased, and we don’t compromise the data quality. Before the analysis, the data were reviewed for unengaged responses, outliers, and data normality. The design of the questionnaire nonetheless permits certain values to be missing.

4.3.1. Data Screening for Careless Responses and Outliers

All the inputs by respondents were checked against outliers and careless responses. We examined participants’ responses to assess their level of attentiveness. The participants’ response patterns were analyzed to identify careless responses, where a respondent might repeatedly select the same response option for consecutive items, and outliers, which may indicate observations that significantly differ from the normal [13]. Within the scope of this study, we measured careless responses by considering both the standard deviation and dimension ratings compared to the average factor ratings. Specifically, we found that two participants consistently provided identical answers, resulting in a standard deviation of zero.

Additionally, the two participants’ group ratings deviated significantly from the average factor ratings within their respective factors. We employed the Mahalanobis distance metric to identify potential multivariate outliers using multiple regression analysis within the SPSS software. In standard units, the Mahalanobis distance measures the squared distance between an observation’s vector and the vector of sample means for all variables [13]. Notably, the probability associated with the Mahalanobis distance fell below 0.001 for four responses. Consequently, these four responses were removed from the dataset [14], leaving us 109 out of the original 113 responses for further analysis.

4.3.2. Normality Test

The normality of the data can be assessed through two main approaches: visual methods (utilizing data histograms, box plots, and Q-Q plots) or numerical techniques (employing measures such as skewness, kurtosis, the Kolmogorov-Smirnov test, or the Shapiro-Wilk normality test). Therefore, the data collected underwent scrutiny for normality distribution using SPSS V26 software. We specifically examined skewness and kurtosis to evaluate the univariate normality of the data for each indicator. As per Pallant [15], skewness reflects the symmetry of the distribution, while kurtosis describes the distribution’s peaks (distribution picks). A skewness and kurtosis value of zero signifies that the data is perfectly normal. However, it’s worth noting that different authors have differing criteria for acceptable skewness and kurtosis values to ensure adherence to normality assumptions, as observed by Byrne [16]. As Kline [17] suggests, absolute values exceeding 3 for skewness and 8 for kurtosis indices indicate severe deviations from normality. Additionally, as per the findings of Xiong et al. [18], skewness and kurtosis values exceeding extreme thresholds serve as clear indications of non-normality. The normality tests for this study show that the absolute skewness values ranged from 0.009 to 2.468 for Indicators Q24 and Q18, respectively. The absolute kurtosis values ranged from 0.057 to 8.915 for factors Q36 to Q18. Both results are not within [18] criteria. Moderate to severe skewness were observed for 6 Indicators (i.e., Q8, Q15, Q17, Q18, Q19 and Q21). Moderate to severe kurtosis were observed for 10 Indicators (i.e., Q8, Q9, Q14, Q15, Q17, Q18, Q19, Q21, Q27 and Q34).

In a Likert-scaled questionnaire, it is common that most respondents select the same scale point, leading to an extremely peaked distribution, resulting in a multivariate positive kurtotic distribution [16]. We employ Maria’s coefficient along with its critical ratio to assess the multivariate kurtosis of the dataset. We employ Maria’s coefficient along with its critical ratio to assess the multivariate kurtosis of the dataset. The dataset is considered to meet the assumption of multivariate normality when the critical ratio (c.r.) falls below 1.96 at a significance level of 0.05. Consequently, the coefficient of multivariate kurtosis approaches nearly zero. A high-value demonstration in Maria’s coefficient suggests of significant positive kurtosis [16]. However, in the present dataset, the z-statistic (c.r.) of 9.85, as depicted in Table 3.2, strongly suggests non-normality within the dataset.

Considering the uncertainty arising from the skewness and kurtosis tests, we further investigated the univariate normality of the data using the Shapiro-Wilk normality test (Ws) within SPSS. As recommended by [19], nonparametric statistical methods are advisable when dealing with fewer than 30 experts or when responses exhibit non-normal distribution, as indicated by skewness. The Shapiro-Wilk Test is particularly well-suited for assessing normality in small sample sizes. A data distribution significantly deviates from normality if the p-value is less than 0.05 [20,21]. The Shapiro-Wilk test (Ws) evaluates the correlation between the provided data and the ideal normal scores. A test value closer to one signifies that the data approximates a normal distribution and supports accepting the null hypothesis, indicating that the data is normally distributed. The formula for the W value is:

$$W=\frac{({\sum }_{i=1}^n{a}_i{x}_{\left(i\right)}{)}^2}{{\sum }_{i=1}^n(x_i-\overline{x}{)}^2}$$

(1)

Where:

$a_i$	constants generated from the covariances, variances, and means of the sample from a normally distributed sample
$x_{\left(i\right)}$	order statistic of a statistical sample
$x_i$	sample values
$n$	sample size
$\overline{x}$	sample mean

As shown in Table 3, The outcome of the Ws test showed that for all indicators, i.e., 24 out of 24 indicators, p-values were consistently below 0.05 and thus is evidence of data non-normality [21,22]. Based on the preceding information, nonparametric estimates are employed in the subsequent sections.

Table 2. Examination of normality of Indicators by the Skewness, Kurtosis values, and Shapiro-Wilk.

Table 2. Examination of normality of Indicators by the Skewness, Kurtosis values, and Shapiro-Wilk.

Code	skew and kurtosis				Shapiro-Wilk
	skew	c.r.	kurtosis	c.r.	Statistic	p-value
I01	-0.808	-3.444	0.131	0.279	0.857	0.000
I02	-0.963	-4.106	1.041	2.218	0.814	0.000
I03	-0.776	-3.308	0.107	0.227	0.836	0.000
I04	-0.471	-2.007	-0.357	-0.761	0.885	0.000
I05	-0.984	-4.196	1.05	2.238	0.771	0.000
I06	-1.234	-5.261	2.252	4.799	0.771	0.000
I07	-1.392	-5.933	2.052	4.372	0.760	0.000
I08	-2.434	-10.375	8.457	18.023	0.609	0.000
I09	-1.144	-4.875	1.185	2.525	0.733	0.000
I10	-1.16	-4.942	1.128	2.404	0.744	0.000
I11	-0.542	-2.31	-0.461	-0.981	0.868	0.000
I12	-0.192	-0.817	-0.443	-0.944	0.855	0.000
I13	-0.009	-0.04	-0.459	-0.979	0.863	0.000
I14	-0.396	-1.689	-0.346	-0.737	0.856	0.000
I15	-0.97	-4.132	1.159	2.47	0.807	0.000
I16	-0.504	-2.15	-0.275	-0.586	0.848	0.000
I17	-0.021	-0.091	-0.382	-0.815	0.879	0.000
I18	-0.624	-2.66	-0.477	-1.017	0.792	0.000
I19	-0.186	-0.795	-1.045	-2.227	0.890	0.000
I20	-0.544	-2.318	-0.255	-0.544	0.877	0.000
I21	-0.471	-2.007	-0.109	-0.232	0.868	0.000
I22	-0.81	-3.453	0.383	0.816	0.794	0.000
I23	-0.261	-1.113	-0.202	-0.431	0.880	0.000
I24	-0.651	-2.776	0.353	0.752	0.853	0.000

4.3.3. Cronbach Alpha

When developing research using Likert Scale data, an important consideration is the questionnaire’s internal consistency. To assess this internal consistency, researchers commonly rely on a widely recommended method, the application of Cronbach’s Alpha coefficient of reliability [15]. In this research, we employ Cronbach’s Alpha coefficient to validate that the Likert Scale measures align with the hypothesis, in line with the Resilience Engineering Indicators (REIs) within the oil and gas sector we intend to assess. Cronbach’s Alpha values range from 0 to 1, “A value of 0.7 is considered acceptable, and 0.8 or higher indicates good internal consistency” [15]. The Cronbach’s Alpha coefficient formula is:

$$\alpha =\frac{\mathrm{k}.\mathrm{r}}{1+\left(\mathrm{k}-1\right).\mathrm{r}}$$

(2)

Where:

$\mathrm{k}$	=	the number of items (factors)
$\mathrm{r}$	=	correlation between the items

The alpha value (∝) increases as the value of k rises. Additionally, a higher Alpha is observed when the items have substantial inter-correlation. The general guideline for Alpha values that typically applies in most situations is as follows:

- The reliability can be considered as excellent when,	0.9 ≤ ∝ ≤ 1.0
- The reliability can be considered as good when,	0.8 ≤ ∝ < 0.9
- The reliability can be considered as acceptable when,	0.7 ≤ ∝ < 0.8
- The reliability can be considered as questionable when,	0.6 ≤ ∝ < 0.7
- The reliability can be considered as poor when,	0.5 ≤ ∝ < 0.6
- The reliability can be considered as unacceptable when,	0.0 ≤ ∝ < 0.5.

We calculated the Cronbach Alpha value for the survey data using the Statistical Package for the Social Sciences (SPSS v.26). The obtained coefficient value, which is 0.869, demonstrates a high level of consistency, as indicated in Table 3.

We employed the reliability coefficient, Cronbach’s alpha (α), from the SPSS package to evaluate the overall scale’s consistency, with a predefined threshold of 0.7 [13]. The reliability analysis results for all variables in this study are presented in Table 3, where all values exceed 0.857. Therefore, the data provided by the respondents exhibit both consistency and reliability, making them suitable for further analysis [21]. Additionally, the alpha value for the entire dataset, measuring 0.869, indicates that the questionnaire scale has achieved acceptable internal consistency and reliability.

5. Results & Discussion

5.2. Ranking Approach

5.2.1. Relative Importance Index (RII)

One of the aims of this study is to gather insights from professionals in the oil and gas industry regarding the most notable REIs. Participants in the survey assessed each REF using a 5-point Likert Scale. The collected data was then analyzed to compute the RII values for each factor. These REIs were subsequently organized in ascending order based on their RII values. To illustrate, a REF with a ranking of 1 signifies the highest level of agreement regarding its impact on organizational Resilience. At the same time, the REF ranked 24th is perceived as the least significant by the participants.

The Relative Importance Index (RII) is employed to establish rankings based on the degree of agreement for each REF. To assess the importance of these resilient indicators and Dimensions, a 5-point Likert Scale was utilized. The RII value falls from 0 to 1, with a higher value signifying greater agreement. The Relative Importance Index has been widely adopted in various studies to rank Dimensions, including several studies in Engineering Management. This method is commonly used to investigate and establish rankings for Indicators based on their relative significance [22,23,24,25]. The authors employed the Relative Importance Index (RII) to assess and arrange the data gathered from the questionnaires. Thus, the calculation of RII can be performed according to the formula presented in Equation 3 below:

$$\mathrm{R}\mathrm{I}\mathrm{I}=\frac{{\sum }_{i=1}^{n}W}{A.N}$$

(3)

Where:

$\mathrm{W}$	=	The weight given to each factor by the respondents (1 to 5)
$A$	=	The highest weight (in this case, the highest weight is 5)
$\mathrm{N}$	=	The total number of responses

RII value ranges from 0 to 1, with a value approaching 1 given more importance than the others lesser than that. According to [26], the RII ranking is as follows:

- The RII can be considered as high when,	0.8 ≤ RII ≤ 1.0
- The RII can be considered as High-Medium when,	0.6 ≤ RII < 0.8
- The RII can be considered as Medium when,	0.4 ≤ RII < 0.6
- The RII can be considered as Medium-Low when,	0.2 ≤ RII < 0.4
- The RII can be considered as Low when,	0.0 ≤ RII < 0.2

Table 5 below displays the RII values and the rankings of the proposed REIs, determined through the agreement scale values provided by all the survey respondents.

The results indicate that the REIs affecting resilience performance have different significance levels, as listed in Table 5. The list of REIs is associated with building organizational capacity (human and systems) and developing the right culture to deal with crises. It is important to note that organizational resilience can be built by effectively implementing these REIs through a competent team working towards achieving the organization’s objectives.

The result of the Relative importance index revealed the relative impact of REIs and its relation to the resilience dimensions, as shown below in Figure 2.

5.2.2. Resilience Performance Index (RPI)

The main finding of this study was that the Resilience Engineering Indicators (REF) could be represented by a single index “Resilience Performance Index (RPI). RPI contains 24 constructs representing the REIs as distributed in 6 dimensions (i.e., D01- Top Management Commitment, D02-Speaking-up Culture, D03-Flexibility, D04-Awareness, D05-Being Prepared, and D06-Learning).

It is important to note that the weight of each resilience factor shall play a role in the overall performance. The Relative importance index would reveal the relative impact of REIs on resilience performance. Therefore, the Resilience Performance Index (RPI) represented the average percentage of the relative importance indices of the 24 indicators. The formula expressed for RPI calculation is shown in Equation 4.

$$\mathrm{R}\mathrm{P}\mathrm{I}=\frac{{\sum }_{i=1}^n{RII}_i}{n}$$

(4)

Where:

$RII$	=	Relative Importance Index of Indicator
$n$	=	Number of Indicators under consideration

The weight of these 24 Indicators is shown below in Figure 3, which shows almost a balanced disruption on the radar chart.

The relative contribution of each indicator to the RPI was calculated according to the formula presented above, and the RPI for the resilience level of the Oil and Gas sector in Qatar is 78.9 % or almost 80 %, which according to Rooshdi et al. [26] can be considered as high level. In other words, Qatar’s Oil and Gas sector showed a high resilience level based on the level of calculated RPI, almost 80%.

Figure 4. Relative effects of RE dimensions.

Figure 4. Relative effects of RE dimensions.

5.2.3. Dimensions Performance Index (DPI)

Following the same concept of RPI as presented in Equation 4, the factor performance by a single index “Dimensions Performance Index (DPI) was calculated as we know that we have five resilience Dimensions namely F01-Top Management Commitment, F02-Speaking-up Culture, F03-Flexibility, F04-Awareness, D05-Being Prepared, and D06-Learning which has 3, 2, 5,5,5,4 REIs respectively as depicted in Figure 5 below.

The resultant FPI scores are listed in Table 6, and Figure 6 shows the FPI score distribution concerning Resilience Dimensions.

6. Conclusion and Recommendations

References

1. M. F. Costella, T. A. Saurin and L. B. d. M. Guimarães, “A method for assessing health and safety management systems from the resilience engineering perspective,” Safety Science, p. 1056–1067, 2009. [CrossRef]
2. J. Wreathall, “Properties of Resilient Organizations: An Initial View,” in Resilience Engineering Concepts and Precepts, Aldershot, Ashgate Publishing Limited, 2006, pp. 275-286.
3. E. Hollnagel and D. D. Woods, JOINT COGNITIVE SYSTEMS Foundations of Cognitive Systems Engineering, Boca Raton, FL: CRC Press Taylor & Francis Group, 2005.
4. World Economics, “World Economics,” April 2023. [Online]. Available: https://www.worldeconomics.com/Wealth/Qatar.aspx.
5. I. G. Union, “13th edition of the IGU World LNG Report,” International Gas Union (IGU), London United Kingdom, 2022.
6. S. Al-Haidous, R. Govindan, A. Elomri and T. Al-Ansari, “An optimization approach to increasing sustainability and enhancing resilience against environmental constraints in LNG supply chains: A Qatar case study,” Energy Reports, p. 9742–9756, 2022. [CrossRef]
7. F. Bento, L. Garotti and M. P. Mercado, “Organizational resilience in the oil and gas industry: A scoping review,” Safety Science, pp. 1-11, 2020. [CrossRef]
8. S. Hosseini, K. Barker and J. E. Ramirez-Marquez, “A review of definitions and measures of system resilience,” Reliability Engineering and System Safety, p. 47–61, 2016. [CrossRef]
9. American Society of Mechanical Engineers (ASME), Innovative Technological Institute (ITI), Washington, DC: ASME ITI, 2009.
10. A. Azadeh, S. M. Asadzadeh and M. Tanhaeean, “A consensus-based AHP for improved assessment of resilience engineering in maintenance organizations,” Journal of Loss Prevention in the Process Industries, pp. 151-160, 2017. [CrossRef]
11. C. Nemeth and R. Cook, “Infusing Healthcare with Resilience,” INCOSE, pp. 1073-1087, 2014. [CrossRef]
12. A. Azadeh, V. Salehi, B. Ashjari and M. Saberi, “Performance evaluation of integrated resilience engineeringfactors by data envelopment analysis: The case of apetrochemical plant,” Process Safety and Environmental Protection, pp. 231-241, 2014. [CrossRef]
13. J. F. Hair, W. C. Black, B. J. Babin and R. E. Anderson, Multivariate data analysis (7th Edition ed.), New Jersey, United States: Pearson, 2014.
14. M. S. Pamulu, “Strategic management practices in the construction industry: a study of Indonesian enterprises,” Queensland University of Technology, Queensland, Australia., 2010.
15. J. Pallant, “Survival manual: A step by step guide to data analysis using SPSS (4th edition ed.),” 2011.
16. B. M. Byrne, Structural equation modeling with AMOS: Basic concepts, applications, and programming, New York: Taylor & Francis Group, 2010.
17. R. B. Kline, “Principles and practice of structural equation modeling,” 2015.
18. B. Xiong, M. Skitmore and B. Xia, “A critical review of structural equation modeling applications in construction research,” Automation in Construction, vol. 49, pp. 59-70, 2015. [CrossRef]
19. S. Kalaian and R. M. Kasim, “Terminating Sequential Delphi Survey Data Collection,” Practical Assessment, Research, and Evaluation, pp. Vol 17, No 5, 2012.
20. A. Field, “Discovering statistics using SPSS (3 ed.),” 2009.
21. H. Zahoor, A. P. C. Chan, R. Gao and W. P. Utama, “The factors contributing to construction accidents in Pakistan: their prioritization using the Delphi technique,” Engineering, Construction and Architectural Management, vol. 24, no. 3, pp. 463-485, 2017. [CrossRef]
22. M. A. Seboru, “An Investigation into Factors Causing Delays in Road Construction Projects in Kenya,” American Journal of Civil Engineering, 2015. [CrossRef]
23. G. A. Bekr, “Factors affecting performance of construction projects in unstable political and economic situations,” ARPN Journal of Engineering and Applied Sciences, vol. 12, no. 19, pp. 5384-5395, 2017.
24. M. Gunduz, Y. Nielsen and M. Ozdemir, “Fuzzy Assessment Model to Estimate the Probability of Delay in Turkish Construction Projects,” Journal of Management in Engineering, vol. 31, no. 4, 2015. [CrossRef]
25. A. L. Sambowo and A. Hidayatno, “Resilience Index Development for the Manufacturing Industry based on Robustness, Resourcefulness, Redundancy, and Rapidity,” International Journal of Technology, pp. 1177-1186, 2021. [CrossRef]
26. R. R. R. M. Rooshdi, M. Z. Abd Majid, S. R. Sahamir and N. A. A. Ismail, “Relative importance index of sustainable design and construction activities criteria for green highway,” Chemical Engineering Transactions, vol. 63, pp. 151-156, 2018.
27. A. P. C. Chan, F. K. W. Wong, C. K. H. Hon, A. Ali Javed and S. Lyu, “Construction safety and health problems of ethnic minority workers in Hong Kong. Engineering, Construction and Architectural Management,” Engineering, Construction and Architectural Management, pp. 901-919, 2017. [CrossRef]
28. A. S. Faridi and S. M. El-Sayegh, “Significant factors causing delay in the UAE construction industry,” Construction management and economics, vol. 24, no. 11, pp. 1167-1176, 2006. [CrossRef]
29. Z. Zamiar and Z. Ścibiorek, “The role of information in crisis management,” Scientific Journal of the Military University of Land Forces, pp. 245-255, 2022.
30. C. Folke, S. R. Carpenter, B. Walker, M. Scheffer, T. Chapin and J. Rockström, “Resilience Thinking: Integrating Resilience, Adaptability and Transformability,” Ecology and Society 15(4): 20, p. 15(4): 20, 2010.
31. Y. Kim, “Building organizational resilience through strategic internal communication and organization–employee relationships,” JOURNAL OF APPLIED COMMUNICATION RESEARCH, p. 589–608, 2021. [CrossRef]
32. S. Margheritti, A. Gragnano, R. Villa, M. Invernizzi, M. Ghetti and M. Miglioretti, “Being an Emotional Business Leader in the Time of the COVID-19 Pandemic: The Importance of Emotions during a Crisis,” Sustainability, pp. 15(4), 3392, 2023. [CrossRef]
33. I. I. PRATIWI, A. APRIANINGSIH, M. Z. AFIF and A. P. PUTRI, “Proposed Model of Business Retail Continuity Process during Pandemic Covid-19 Based on Risk identification and Response,” CEEOL, pp. 12-07, 2021.
34. E. HOLLNAGEL, D. D. WOODS and N. LEVESON, Resilience Engineering Concepts and Precepts, Aldershot, UK: Ashgate Publishing Limited, 2006.
35. S. Nassereddine, “Corporate governance between the theoretical framework and application mechanisms: literature review,” Journal of Financial, Accounting and Managerial Studies, pp. 337-358, 2022.
36. A. Yahya, “Analysis of project success factors in Middle east construction industry,” American University of Sharjah, 2014.
37. M. Li, “A waste management system for small and medium enterprises engaged in office building retrofit projects. (Doctor of Philosophy),” Queensland University of Technology, Queensland, Australia, 2012.