For Dr. Frank only please see the attached.

CHAPTER 3: METHODOLOGY 46

Introduction 46

Research methodology and design 47

Population and sample 48

Data sources/instruments/ materials 51

Reliability and Validity of Instruments 52

Data collection procedures 53

Data analysis procedures 54

Assumptions 57

Limitations 57

Delimitations 58

Ethical assurances 58

Internal and external Validity 58

Summary 59

CHAPTER 4: RESULTS, ANALYSIS AND FINDINGS 61

Data Collection and Screening 62

Description of the Sample 63

Study Measures 65

Quality Criteria: Reliability Test 65

Pearson Correlations 68

Normality Test using Histogram 68

Scatter Plot 77

Data Analysis 88

Correlation 89

Regression 92

Summary 93

Considering such a huge population size, correspondingly larger sample size is required. However, considerably larger sample size has undesirable logistical implications. For instance, going with the standard definition of a representative sample size as illustrated by Cochran (2007), 10 percent of the total population would be required. Collecting data from such a huge sample size would consume a considerable amount of time. An alternative approach to the determination of appropriate sample size is Cochran's (2007) G*power technique. The G*power technique does not require estimation can be used for situations in which the population size is either unknown or extremely large hence not feasible for effective sampling.

A priori power analysis was conducted using G*Power to determine the required minimum sample size for the study. Four factors were considered in the power analysis: significance level, effect size, the power of the test, and statistical technique. The significance level, also known as Type I error, refers to the chance of rejecting a null hypothesis given that it is true (Haas, 2012). Most quantitative studies make use of a 95 percent confidence level because it adequately provides enough statistical evidence of a test (Creswell & Poth, 2017). The effect size refers to the estimated measurement of the relationship between the variables being considered (Cohen, 1988). Cohen (1988) categorizes effect size into small, medium, and large. Berger, Bayarri, and Pericchi (2013) purported that a medium effect size is better as it strikes a balance between being too strict (small) and too lenient (large). The power of the test refers to the probability of correctly rejecting a null hypothesis (Sullivan & Feinn, 2012). In most quantitative studies, an 80 percent power is usually used (Sullivan, & Feinn, 2012).

The statistical test to be used for this study is multiple regression. In order to conduct multiple regression to detect a medium effect size, at the 5 percent level of significance, with 80 percent power, at least 109 participants are required (Figure 1). This calculation is also based on eight independent variables: leadership type (transformational leadership, transactional leadership, passive-avoidant behaviors) and toxic leadership (self-promotion, abusive supervision, unpredictability, narcissism, and authoritarian leadership). Additionally, gender and age will be controlled for in the analysis.

Figure 1 G*Power Sample Size Estimation for Multiple Regression

The comment is below:

I think it looks fantastic. You have figure 1 on page 50 which appears with no mention of where or how you obtained it. If you generated it, make that clear or if you used a product, report that. I used Qualtrics to get this information. Do I have to change my whole table of content? Is there an easier way? I don’t want to mess up my format.