98 CHAPTER 3 METHODOLOGY 3.1 Introduction This chapter discusses the methodology and procedures followed in conducting this study. It includes the research approach and design. Also, it identifies the location of the study, the population, sample size and research instruments. Then, it discusses data collection and procedures. Finally, it presents the statistical methods that has been used to analyse data and get the results. 3.2 Research Design The underlying framework for integrating all aspects of a quantitative study such that the results are trustworthy, devoid of bias and maximum generalizable is known as research design. The study design defines how participants are chosen, which variables are included, how they are modified, how data is gathered and evaluated and how unnecessary variability is managed in order to answer the main research topic (Venters et al., 2018). The selection of appropriate inquiry techniques can describe the study design. Moreover, Creswell (2017) summarized the quantitative research method as mentioned in Table 3.1. Table 3.1: Quantitative Research Method Quantitative Methods It is a predetermined method. Data instruments are based on questions. It includes performance data, attitude data, observational data and data census. The analysis is based on statistical methods. The interpretation is based on statistical methods Source: Creswell (2017). 99 Applied research is a methodology used to address a complex social issue. The research can be categorized as quantitative or qualitative based on the type of data that is used. There are three main categories of research: experimental research, hypothesis testing and exploratory research, to investigate a non-clearly defined problem. It is to implement a search for understanding the current problem better and offer indicative, but not necessary, final results (Garagorri, 2016). These preliminary findings may be used to develop new research questions or hypotheses for further investigation (Teshome, 2019). Quantitative research depends on gathering and analyzing numerical data to define, illustrate forecast, or monitor variables and phenomena of interest (Gay et al., 2012). This study applies an individual basis as the unit of analysis. The collected data were analyzed using SPSS to define the factors, which influence the motivation and how they affect their productivity. At the same time, Smart PLS has also been applied to evaluate the measurement model and to analyze the study's mediator variable. Based on Creswell (2017), the quantitative design is appropriate for a study where tendencies or clarifications required to be achieved. Moreover, a quantitative research design empowers to supply accurate and objective descriptions of the matter being considered (Taylor et al., 2017). A quantitative research design allows analysts to handles the data by collecting numerical data, then changing it to figures and numbers to extricate the meaning. Also, it tends to utilize agent tests related to more significant studies and the data picked up from the agent sample permits analysts to conclude almost the general research population (Bryman and Chime, 2007; Martyn, 2010). In the current study, quantitative analysis employed to investigate the influence of the leadership role and the motivational factors on the OWS employees’ productivity. Also, to prove the research hypotheses using the quantitative 100 method for data collection. For answering the study questions, the online questionnaire technique was used to collect data from the sample size. In this study, the research questions helped in designing the model, answer research questions and the responses helped to evaluate research hypotheses. The following table 3.2 represents the research design and approach. Table 3.2: Research design and approach Title Title Method Research design Exploratory research design based on a quantitative study. Research approach Quantitative approach. Research strategy The survey used a questionnaire. Sample selection Stratified Random Sampling applied for this research. Data collection A questionnaire used for data collection. Data analysis Statistical analysis using SPSS and Smart PLS. Figure 3.1 illustrates the framework of data collection and data analysis, which have been used for the present study. An instrument of the questionnaire has been adopted for data collection aiming to answer the research questions. The population of the current study includes all levels of employees in the Oman Water Sector. Therefore, a stratified random method has been applied for gathering data from participants. Then, SPSS is applied to test the primary analysis and Smart PLS to evaluate the measurement model, structural model and mediator variable analysis of the study. Based on the findings, conclusions and recommendations of research have been written. 101 Figure 3.1: Data Collection and Data Analysis Framework 3.3 Research Philosophy Research philosophy refers to a set of ideas about how knowledge. It develops on the basis of certain assumptions that describe how the researcher sees the world, which is essential in research philosophy. The first sort of research philosophy is positivism, in which the researcher develops hypotheses to test the study aims (Saunders et al. 2012). In positivist Research Questions Data Collection (Questionnaire) Sampling Method (Stratified Random Sampling Method) Participants Population (Employees of Water Sector in Oman) Data Analysis (SPSS & Smart PLS) Sample size (Employees of all levels in OWS) Findings, Discussion, Conclusion and Recommendations 102 philosophy, the data collection approach is generally quantitative in nature and highly organized (Sardar, 2017). A research philosophy is a set of ideas and assumptions regarding the evolution of knowledge. Although this may seem profound, it is precisely what a person does when beginning on research, to develop knowledge in a certain subject. A person's knowledge development may not be as spectacular as inventing a new theory of human motivation, but even solving a specific problem in a specific organization requires the creation of new knowledge (Saunders et al., 2015). Positivism is a natural scientist's philosophical position that includes working with observable social reality to generate law- like generalizations. In other terms, positivism refers to the natural scientist's philosophical perspective. Dealing with an observable social reality is required, and the final result can be law-like assumptions comparable to those found in the physical and natural sciences (Saunders et al., 2015). The current study applies the positivism research philosophy because it adopts a deductive approach to find the relationship between the employees’ productivity and the leadership role in the workplace. Moreover, the main research inquiry will be answered statistically through the research questions. For quantitative data, a descriptive analysis and discussion were performed to ensure a thorough understanding and to provide the research inquiry's answers in a structured way. 3.4 Research Method (Stratified Random Sampling) This study used Stratified Random Sampling (SRS) to highlight specific subgroups within the research's target population and find their relationship. In addition, SRS used to estimate the population parameters for groups within the population. Stratified Random Sampling 103 (SRS) is a generally utilized testing method for approximate query processing (Nguyen et al., 2019). The data is partitioned into different sub-gatherings (layers) based on basic characteristics such as age, sex, ethnicity, income, education and culture. Any layer receives a random sample. Every stratum's characteristic can be measured and distinctions can be made. Also, any layer's characteristics can be measured, and comparisons can be made. It also reduces the uncertainty caused by systematic sampling. The disadvantages are that it necessarily requires accurate information from each layer's proportions, as well as the fact that stratified lists are expensive to create (Acharya et al., 2013). In sample collection, auxiliary variables are often used to increase the accuracy of estimates. At whatever point the reis auxiliary the data accessible, the researchers need to use it in the estimation technique to get the most proficient estimator. A standard variance, coefficient of difference, skewness, kurtosis and correlation coefficient, among other auxiliary variables, can be used to identify an auxiliary variable and distinct parameters (Keshavarzi et al., 2021). Many scholars, such as Sisodia and Dwivedi (1981), Upadhyaya and Singh (1999), and Singh and Tailor (2003) have developed various estimators to boost the ratio estimators in the simpler and sampling cases. The estimators in Upadhyaya and Singh (1999) were adapted to Stratified Random Sampling by Kadilar and Cingi (2003). In stratified random sampling, Singh and Vishwakarma (2008) suggested a new set of estimators. Castro et al. (2017) proposed a stratified random sampling method to deal with the division of populations in all player phases into sub-populations with equal sizes of predecessors for each player in order to increase estimation accuracy. A procedure for determining each stratum's sample size must be developed to use a stratified random sampling process. Castro et al. (2009) proposed a two-phase Shapley value estimation 104 algorithm with optimal sample distribution, using real variance as a metric to distribute samples across layers to reduce the expected error. To achieve an initial approximate Shapley Value and having each stratum's sample variance, half of the samples are uniformly allocated to each stratum at the first step. Then, the remaining half of the samples are optimally allocated to each layer based on the sample changes chosen in the first step. The final approximate Shapley value is then calculated based on the two phases' sampling effects (Han et al., 2019). The collection of Voxels, where independent reference data was gathered, was a sampling issue and a stratified random sampling design was proposed in an analysis for Stehman (2009). Where the cost of sampling is minimal and simple random sampling is sufficient. However, only a small sample of independent reference data can be obtained for object validation. An interpreter's requirement to either generate the reference data through visual representation or calculate and correct the effects of a semi-automatic process can also be gathered (Boschetti et al., 2016). When an independent reference data sampling design is used instead of simple random sampling, the sample size is similar. The standard errors of the precision and area estimators are smaller (Stehman 1997). Following that, stratified random sampling was introduced as a viable option for item validation (Boschetti et al., 2006). 3.5 Location of the Study The current study is implemented in the Sultanate of Oman and focuses on the water sector, which provides potable water services for the customers. The population includes Ministry of Agriculture, Fisheries and Water Resource (MAFWR) and Public Authority for Water (PAW). This research study is limited to the mentioned organizations only because of its 105 importance for public services to the audience and customers. Expanding the motivation in such a sector will result, at last, in the work execution and quality towards those customers with high fulfillment. Furthermore, the study has used the mentioned organizations due to the lack of empirical research in such sector organizations. In this research, the study population comprised all the MAFWR and PAW staff levels of the Oman Water Sector as detailed in Table no. 3.3 and Table no. 3.4. 3.6 Population and Sampling This study has a population of 2264 employees in the OWS, which represents 453 employees in MAFWR and 1811employees in PAW (MAFWR, 2020 & PAW, 2020). The details of the sample size have been classified in the following Table no. 3.5 and Table no. 3.6. Defining this sample size was based on the Krejcie and Morgan Table for Sample Size Determination of a Given Population (1970). This calculation with respect to confidence level 95% and a margin of error of 5%. Stratified Random Sampling (SRS) is a quantitative method used by the researcher for this study. For data collection from the targeted population and a questionnaire has been used as the research instrument. Lee Chuan (2006) utilized Krejcie and Morgan, a prominent research method, to estimate sample size in his research. The below formula was used by Krejcie and Morgan (1970) to determine sample size: S = X2 NP (1-P) / d2 (N-1) + X2P (1-P) S: needs sample size, X2: refers to the chi-square table value for one-degree freedom at the desired level of confidence, 106 N: represent the population size, P: refers to the proportion of the population (assumed to be 0.50 since this would provide the maximum sample size), and D: refers to the accuracy degree expressed as a proportion (0.05). In the following three tables present the detailed figures of the employees' positions and the sample size of this study, which includes the managerial and non-managerial positions in the Oman Water sector: the Ministry of Agriculture, Fisheries and Water Resource (MAFWR) and the Public Authority for Water (PAW). Table 3.3: Employees' Positions in PAW Position Title No. General Manager 6 Senior Manager 14 Department Manager 67 Section Head 141 Non-managerial positions 1583 Total of all employees 1811 Table 3.4: Employees' Positions in MAFWR (Water Sector) Position Title No. General Manager 2 Assistant of General Manager 4 Department Manager 17 Assistant of Department Manager 17 Section Head 79 Non-managerial positions 334 Total of all employees 453 107 Table 3.5: Determining Sample Size of Study Statement Figure Population in PAW 1811 Population in MAFWR (Water Sector only) 453 Total Population in OWS (PAW & MAFWR) 2264 The Margin of Error (Confidence Interval) MoE = 5 % Significance (Confidence) Level Confidence = 95 Percentage of population in PAW (1811/ 2264) X 100 = 80 % Percentage of population in MAFWR (453 / 2264) X 100 = 20 % Sample Size in PAW 329 X 80 % = 263 Sample Size in MAFWR 329 X 20 % = 66 Total of Sample Size OWS (PAW & MAFWR) 329 Based on table 3.7, Krejcie and Morgan (1970), a detailed population and the related sample size have mentioned in the above table. The sample size has calculated based on the Margin of Error of 5 % and Significance Level with 95 confidences. 108 Table 3.6: Table for Determining Sample Size from a Given Population Source: Krejcie and Morgan (1970). 109 There are some steps to apply and perform a stratified sampling of this study. They are starting by dividing the population into smaller sub-groups or strata based on the population's organizations and then based on their types of employees' positions. Then, taking the random sample in an equal amount to the stratum's size from each stratum. Next, the subsets of the strata are pooled together to form a random sample. Finally, data has been collected from respondents and conducting the data analysis. The detail of determining the sample size according to the position types is detailed, as shown in Table No. 3.7. Table 3.7: Determining Sample in OWS Based on Position Types Position Title No. of Employees % of Employees' Total Sample Size General Manager 8 0.35 % 0.0035 X 329 = 1.152 Assistant of General Manager/ Senior Manager 18 0.8 % 0.008 X 329 = 2.632 Department Manager 84 3.7 % 0.037 X 329 = 12.173 Assistant of Department Manager 17 0.75 % 0.0075 X 329 = 2.468 Section Head 220 9.7 % 0.097 X 329 = 31.913 Non-managerial positions 1917 84.7 % 0.847 X 329 = 278.663 Total of all employees 2264 100 % Total of sample size = 329 110 Table 3.8: Detailed Sample Size Based on Position in PAW and MAFWR Position Title No. of Employees in PAW No. of Employees in MAFWR Sample Size in PAW Sample Size in MAFWR General Manager 6 2 0.003 X 263 = 0.789 0.004 X 66 = 0.264 Assistant of General Manager/ Senior Manager 14 4 0.008 X 263 = 2.104 0.009 X 66 = 0.594 Department Manager 67 17 0.037 X 263 = 9.731 0.038 X 66 = 2.508 Assistant of Department Manager 0 17 0 0.038 X 66 = 2.508 Section Head 141 79 0.078 X 263 = 20.514 0.174 X 66 = 11.484 Non-managerial positions 1583 334 0.874 X 263 = 229.862 0.737 X 66 = 48.642 Total of all employees 1811 453 263 66 Grand Total 2264 329 3.7 Research Instruments This study depends on distributing and analyzing specific questionnaire for gathering data and information that supports doing this academic research. Also, linked information from related literature of associated studies has been used to expose what other researchers found in a similar field. The questionnaire for this study has been designed from two sections as reviewed from previous studies. Developing this instrument is based on the studies of Njambi (2014), Alaidarous and Al-Mahdhoori (2015), Al-Manthri and Correia-Harker 111 (2016), Gabriela and Dorinela (2017) and Alfagira (2019). The first section includes 56 questions distributed in three dimensions. The first dimension is about the leadership role comprises of 13 statements. The second dimension is about motivation which includes 19 statements. The third dimension is about employee productivity that includes 21 statements. The questions in this section were created based on the assumption that such variables directly influence leadership's role. The second section is about demographic information for participants such as sex, age, education level, position and work experience. The Likert Scale of a five-point rating system was adopted to reflect the intensity of respondents’ opinions. It gives some degree of choice and simplifies mathematical interpretation. The Likert Scale was used to explore the participants' perceptions about the leadership role, where point "5" represents strongly agree and "1" strongly disagrees (Boone, 2012). Table 3.9: Sources of the Research Instrument Section / Dimension Source The Leadership Role Gabriela & Dorinela, 2017; Alfagira, 2019 Motivation Factors Njambi (2014) Employees Productivity Alaidarous (2015) Demographic Information about Participants Al-Mahdhoori, 2015; Al-Manthri & Correia-Harker, 2016 3.8 Pilot Study One of the advantages of doing a pilot study is that it suggests areas where the main research effort might fail, where research procedures might not be implemented or where potential approaches or instruments might be ineffective or complicated (Bhagwat-Chitale, 2017). Large-scale surveys can use pilot studies before the primary survey is performed and pilot 112 studies can depend on qualitative and quantitative approaches. As a result, it might begin with qualitative data collection and interpretation of comparatively unexplored research subjects, then use the findings to plan a quantitative phase of the research (Johnson & Walsh, 2019). To ensure the questionnaire's reliability and feasibility, the researcher performed a pilot study, as detailed in the following topics, with a group of participants who were close to the real participants of the sample size. The number of participants who responded to the questionnaire of the pilot study was 46 responses. Random sampling was selected and every element in the population is considered and has taken as a subject (Sekaran et al., 2019). A pilot study with a group of participants was carried out in a similar way to the real participants of this study. It is to ensure the questionnaire reliability and validity study. When random sampling is the procedure, a pilot study's sample size can be sensibly run from 30 to 50 (Hertzog, 2008). Accordingly, the number of participants who responded to the pilot study was 46 respondents. The researcher has designed the research questionnaire by using Google Forms. This technique has some advantages. The researcher used this method of Google Forms because the respondents will answer the questions mandatorily, which ensures there are no missing questions. It is easier to be distributed and then collected automatically by emails. Also, the collected data are more organized and easier for statistical analysis. The reliability of the questionnaire's research instrument developed by Correia- Harker (2016) has found that reliability with Cronbach’s Alpha of 0.87 for leadership and motivation has a good reliability level with Cronbach’s alpha 0.83. Also, Al-Mahdhoori (2015) has found that his instrument's reliability is ranged from 0.75 to 0.92, which shows high internal consistency. Whereas, Njambi (2014) has found that the Coefficient Alpha 113 Reliability 0.809 for extrinsic factors of motivation and 0.861 for intrinsic factors. Alaidarous (2015) has found that reliability with Cronbach's alpha of 0.848. Table 3.10: Reliability of original instrument Scholar Reliability of Cronbach’s Alpha Correia-Harker (2016) 0.87 Correia-Harker (2016) 0.83 Al-Mahdhoori (2015) 0.75 to 0.92 Njambi (2014) 0.809 to 0.861 Alaidarous (2015) 0.848 Through using the SPSS version 24.0, all scales' reliability was measured using an internal consistency indicator, namely Cronbach’s Alpha, which indicated that all constructs are reliable. The below Table 3.11 shows that the reliability from 0.947 to 0.974 for 53 questions of the research instrument. The first dimension of leadership role has got 0.951 of Cronbach’s Alpha. The second dimension includes the motivation that got 0.947 of Cronbach’s Alpha. Also, the third dimension is employees' productivity, with 0.974 of Cronbach’s Alpha. Table 3.11: Reliability Test of Pilot Study Dimension Cronbach’s Alpha (α) The Leadership Role 0.951 Motivation 0.947 Employees Productivity 0.974 Hair et al. (2017) recommended that the estimated value of reliability that scored 0.7 or higher is considered a good value. Therefore, the statistical finding of a pilot study 114 shows a high rate of Cronbach's Alpha. The researcher explores the possibility of using the same questionnaire as valid for application as a research tool. The detailed statistical analysis data provided by SPSS version 24.0 are in the following appendices no. 6, 7, 8, and appendix 9. 3.9 Ethical Considerations in Data Collection While gathering data that involves humans, there are various standards need to be followed. Additionally, boards and systems to monitor the process are generally in place (Schuller et al., 2016). The researcher has taken the following steps that aim at ethical consideration of data collection: i. The researcher has officially requested the organizations, which are included in the scope of the study to collect data from their employees. ii. The data collection instrument that was used in the survey for this study has conformed to the respondents' confidentiality about the confidentiality of gathered data. It ensured that the data used would only be used for academic purposes. iii. The researcher did not request the name of the respondent to keep the personality data confidential. iv. The collected data was not shared with any person or any open source. v. Access to the gathered data was limited to the researcher only. 3.10 Reliability and Validity Reliability is the degree to which results are constant over time and how the whole sample population is accurately represented (Golafshani, 2013). While Kirk and Miller (1986) 115 have classified the level at which a measurement stays the same, the overtime measurement stability and the similarity between measures within a particular time frame. While the three forms of reliability applicable to quantitative analysis are classified. Charles (2013), on the other hand, adheres to the idea that it is possible to ascertain in two different circumstances the accuracy with which the questionnaire items are answered in the same way. Cronbach's Alpha is the most popular tool for calculating reliability. It mentioned that an acceptable situation of reliability should be above 0.7, but if the Cronbach's Alpha has been found below 0.7, it means low internal consistency in the variable or lower reliability (Pallant, 2010). Cronbach's Alpha allows the internal consistency for several items to belong together and it shares similarities in their measurements of a specific construct (Sashkin & Sashkin, 2003). Many researchers have discussed the concepts of validity and they developed their concepts where have often generating or adopting what they believe to be more suitable terminology, such as rigor, consistency and trustworthiness (Davies & Dodd, 2002). Therefore, this study's questionnaire is going through the validation process to ensure its suitability and quality. The construct validity of the questionnaire's research instrument developed by Correia-Harker (2016) has been tested by investigating how well the leadership motivation scale was held together and whether the scale was a particular yet interconnected build-in connection between leadership self- efficacy and leadership capacity. For this study, the research instrument has exposed 13 arbitrators for the arbitration process (Appendix No. 5). They provided valued comments in developing the questionnaire in the best manner to be suitable for collecting data from the participants. Their comments and advice included spilling and correcting the clauses, modifying, deleting, adding some 116 statements and transferring some statements from one dimension to another for more suitability. Then, the questionnaire has translated into the Arabic language to be clearer and easier to understand by the respondents. Three arbitrators have reviewed and arbitrated the translation from Arabic to English. Theses arbitrators are academic people who well know about the Arabic and English language. Their comments and suggestions included spilling and revising the clauses, as well as changing, removing and adding certain words in the statements to make them more suitable. This process was to ensure the quality and clarity of the questionnaire before distribution to the participant. 3.11 Data Collection and Procedure The data collection method is a process used for data collection based on the required data. It includes the investigation of research questions that need to be answered. Also, it identifies the data type for each question. Such as nominal, ordinal, interval or ratio. Moreover, it determines the sample unit's features to gain the participants' thoughts, ideas and experiences (Cooper & Schindler, 2011). This research has consecrated to the usage of primary data from the target sample obtained. The data was gathered using a formal questionnaire. The data collection tool in this research has been established based on the literature of numerous scholars on related subjects that affect workers' morale and their effect on their organization’s productivity. Because the questionnaire has spread over a large area of Oman, it becomes the best method for collecting the required data and it needs to cover many of the widely scattered respondents. To start collecting all data from a population, the researcher has officially requested MAFWR and PAW (see Appendix No. 1 and 2). The data was collected by using 117 Google Forms where the research questionnaire was designed. Then, the researcher sends emails to the participants and follows up with their responses. The data was received and collected automatically by Google Forms. All these data have been collected from the respondents during the period of 1st to 30th June 2020. 3.12 Chapter Summary This chapter outlined the academic study's research methodology, design, data collection, instrument, and data analysis process. The data has obtained by the use of a questionnaire instrument. Before using the questionnaire in data gathering from respondents, academic people have arbitrated it for the developing purpose. A pilot study has then implemented before distributing the final draft of the questionnaire sheet for the representation. After gathering data, it has gone into the process of testing or maximizing its validity and reliability. The IBM SPSS software platform Version 24 has been used to analyze the gathered data used for interactive or batched statistical analysis.