CHAPTER FOUR: FINDINGS
Overview
The purpose of this quantitative, causal-comparative study was to compare the main effect of gender (female/male) and level of religiosity (higher/lower) on the prevalence of cyberbullying experiences, as a victim and an offender, among traditional undergraduate students attending two faith-based universities in the southern United States during the 2020 fall semester. The independent variables are biological gender (female/male) and level of religiosity (higher/lower). The dependent variables are the prevalence of cyberbullying victimization experiences and the prevalence of cyberbullying offending experiences. This chapter includes a review of the study’s research questions, followed by the six null hypotheses associated with the two research questions. In addition, the chapter reports the descriptive statistics and a complete analysis of all data collected.
Research Questions
RQ1: Is there a difference in the prevalence of cyberbullying victimization experiences, based on biological gender and level of religiosity, among traditional undergraduate students attending faith-based universities?
RQ2: Is there a difference in the prevalence of cyberbullying offending experiences, based on biological gender and level of religiosity, among traditional undergraduate students attending faith-based universities?
Null Hypotheses
Ho1: There is no difference in the prevalence of cyberbullying victimization experiences scores between female and male traditional undergraduate students attending faith-based universities.
Ho2: There is no difference in the prevalence of cyberbullying victimization experiences scores, based on level of religiosity (higher/lower), of traditional undergraduate students attending faith-based universities.
Ho3: There is no difference in the prevalence of cyberbullying victimization experiences scores, based on biological gender (female/male) and level of religiosity (higher/lower), of traditional undergraduate students attending faith-based universities.
Ho4: There is no difference in the prevalence of cyberbullying offending experiences scores between female and male traditional undergraduate students attending faith-based universities.
Ho5: There is no difference in the prevalence of cyberbullying offending experiences scores, based on level of religiosity (higher/lower), of traditional undergraduate students attending faith-based universities.
Ho6: There is no difference in the prevalence of cyberbullying offending experiences scores, based on biological gender (female/male) and level of religiosity (higher/lower), of traditional undergraduate students attending faith-based universities.
Descriptive Statistics
All respondents completed an anonymous, online survey using QualtricsXM that included four separately scored sections. The COAS instrument included 10 questions on cyberbullying victimization and 10 questions on cyberbullying aggression (offending). Each question was scored on a 3-point Likert-type scale ranging from 0 (never) to 3 (many times). The highest score for each section is 30, indicating the respondent had been a cyberbullying victim or offender many times. The lowest score for each section is 0, indicating the respondent reported never being a cyberbullying victim or offender. According to Warner (2013), reliability is the consistency of measurement of results; therefore, a Cronbach’s alpha coefficient was used to calculate reliability for each instrument, with α = > 0.70 indicating a strong reliability. The COAS victimization scale for the participating respondents (N = 284) had an acceptable Cronbach’s alpha coefficient with α = .708. The COAS offending scale also had an acceptable Cronbach’s alpha coefficient with α = .720.
The third part of the survey was the DUREL, which measured the respondent’s level of religiosity. The DUREL utilizes five questions that are scored on a 6-point Likert-type scale ranging from 1 (more than once a week) to 6 (never) for Questions 1 and 2 and a 5-point Likerttype scale for ranging from 1 (definitely true of me) to 6 (rarely or never) Questions 3 through 5. Cronbach’s alpha coefficient for the DUREL was α = .857, indicating a high internal consistency for the instrument. Finally, the survey included a researcher-created demographic question to collect data on the respondent’s biological gender. To analyze the data, the following dummy codes were used: females were coded as 1, males were coded as 2, and those who preferred not to answer were coded as 3.
The research sample includes 284 (N = 284) students from both participating universities who submitted completed surveys with all information needed. Three hundred and twenty-four responses were received; however, 40 were incomplete or did not contain sufficient information to be included in the study. The majority of the respondents (218 or 76.7%) were traditional undergraduate students at University A, while the remaining 66 respondents (23.3%) were traditional undergraduate students at University B. The sample includes 180 female students (62.7%) and 104 male students (36.3%). The raw scores on the DUREL ranged from 5 to 27 (M = 10.51, SD = 4.71). Each respondent’s raw scores on the DUREL were calculated, and a level of religiosity was assigned. Of the 284 participants (N = 284), 209 (73.9%) scored a 13 or below, indicating a higher level of religiosity, and 75 (26.1%) scored a 14 or above, indicating a lower level of religiosity. Descriptive statistics are provided for cyberbullying victimization by gender and level of religiosity (see Table 1).
Table 1 Descriptive Statistics: Cyberbullying Victimization
Descriptive statistics were also calculated for cyberbullying offending experiences based on the two independent variables of gender (female/male) and level of religiosity (higher/lower) with a composite total score (see Table 2).
Table 2 Descriptive Statistics: Cyberbullying Offending Scale
Results
Two, two-way analyses of variance (ANOVA) were utilized to compare total scores on the COAS victimization and COAS offending scales, respectively, with two categorical independent variables. The two categorical independent variables were gender (female or male) and level of religiosity (higher or lower). The following section includes a description of the assumption testing utilized to ensure all data met the necessary assumptions for each two-way ANOVA. An analysis of each research hypothesis is also included in this section.
Data Screening
To test for extreme outliers for the dependent variables of cyberbullying victimization and cyberbullying offending, four separate box-and-whisker plots were used (see Figures 1–4). Extreme outliers were detected in all four plots; however, the researcher made a judgment call to include the extreme outliers because they are important to the overall outcome of the study. According to Warner (2013), it is a researcher’s decision to include, or not include, outliers based on the study and the sample size. Two extreme outliers (7, 117) were found for female cyberbullying victimization experiences scores, and one extreme outlier (273) was found for male cyberbullying victimization experiences scores. Two extreme outliers (7, 117) were found for cyberbullying victimization experiences of students with higher levels of religiosity, and one extreme outlier (273) was found for students with lower levels of religiosity. Four extreme outliers (137, 138, 140, and 150) were found for female cyberbullying offending experiences scores, and five extreme outliers (205, 256, 244, 267, and 269) were found for male cyberbullying offending experiences scores. Four extreme outliers (199, 205, 234, and 244) were found for students who were cyberbullying offenders with higher levels of religiosity, and four extreme outliers (138, 163, 267, and 269) were found for cyberbullying offenders with lower levels of religiosity.
Figure 1. Box-and-whisker plot for cyberbullying victimization and gender.
Figure 2. Box-and-whisker plot for cyberbullying victimization and level of religiosity.
Figure 3. Box-and-whisker plot for cyberbullying offending and gender.
Figure 4. Box-and-whisker plot for cyberbullying offender and level of religiosity.
Assumptions
The assumption of normality was tested using the Kolmogorov-Smirnov test for both independent variables, and the results were significant for all groups; therefore, the null hypotheses were rejected for the assumption of a normal distribution across gender and level of religiosity (see Tables 3 and 4). Although this assumption was not tenable, the two-way ANOVA analysis is robust enough to handle violations of this assumption (Warner, 2013).
Levene’s test of equality of error variances was used to determine homogeneity of variance for both scales. A two-way ANOVA assumes for the dependent variable that the population variances are equal for all groups of the independent variables. For cyberbullying victimization, the test was not significant, F (3, 280) = 1.20, p = 0.31. For cyberbullying offending, the test was not significant, F (3, 280) = 2.27, p = 0.08. Therefore, the assumption of variances was found to be tenable for this study. Due to running two tests of significance, a Bonferroni correction was used to guard against Type I error. The alpha level was calculated as: PCα = EWα/k where, PCα is the experiment-wise α, typically α =.05 and k is the number of significance tests performed, which is 2 in this study. Therefore, PCα = .05/2, PCα = .025, and rounded to α =.03.
Table 3 Tests of Normality: Cyberbullying Victimization Scale
Table 4 Tests of Normality: Cyberbullying Offending Scale
Results for Null Hypothesis One
A two-way ANOVA was used to test the first null hypothesis regarding differences in the prevalence of cyberbullying victimization experiences scores between female and male traditional undergraduate students attending faith-based universities (see Table 5). The null hypothesis was not rejected at a 95% confidence level where F(1, 280) = .215, p = .643, n2 = .001. The effect size was small. There was no significant difference in the prevalence of cyberbullying victimization experiences scores between traditional undergraduate female students (M = 1.25, SD = 0.16) and traditional undergraduate male students (M = 1.13, SD = 0.20) attending faith-based universities.
Results for Null Hypothesis Two
A two-way ANOVA was used to test the second null hypothesis regarding differences in the prevalence of cyberbullying victimization experiences scores based on level of religiosity (higher/lower) among traditional undergraduate students attending faith-based universities (see Table 5). The null hypothesis failed to be rejected at a 95% confidence level where F(1, 280) = 1.88, p = .172, n2 = .007. The effect size was small. There was no significant difference in the prevalence of cyberbullying victimization experiences scores between traditional undergraduate students with a higher level of religiosity (M = 1.01, SD = 0.13) and traditional undergraduate students with a lower level of religiosity (M = 1.36, SD = 0.22) who attend faith-based universities.
Results for Null Hypothesis Three
A two-way ANOVA was used to test the third null hypothesis regarding the interaction of prevalence of cyberbullying victimization experiences scores based on biological gender (female/male) and level of religiosity (higher/lower) among traditional undergraduate students attending faith-based universities (see Table 5). The null hypothesis failed to be rejected at a 95% confidence level where F(1, 280) = 1.23, p = .267, n2 = .004. The effect size was small. There was no significant difference between the prevalence of cyberbullying victimization experiences scores for traditional undergraduate female students with a higher level of religiosity (M = 1.22, SD = 2.18) and the prevalence of cyberbullying victimization experiences scores for traditional undergraduate male students with a higher level of religiosity (M = 0.81, SD = 1.12) who attend faith-based universities. There was also no significant difference between the prevalence of cyberbullying victimization experiences scores for traditional undergraduate female students with a lower level of religiosity (M = 1.28, SD = 1.36) and the prevalence of cyberbullying victimization experiences scores for traditional undergraduate male students with a lower level of religiosity (M = 1.45, SD = 2.24) who attend faith-based universities.
Results for Null Hypothesis Four
A two-way ANOVA was used to test the fourth null hypothesis regarding differences in the prevalence of cyberbullying offending experiences scores between female and male traditional undergraduate students attending faith-based universities (see Table 6). The null hypothesis failed to be rejected at a 95% confidence level where F(1, 280) = 3.10, p = .080, n2 = .011. The effect size was small. There was no significant difference in the prevalence of cyberbullying offending experiences scores between traditional undergraduate female students (M = 0.28, SD = 0.08) and traditional undergraduate male students (M = 0.49, SD = 0.10) who attend faith-based universities.
Results for Null Hypothesis Five
A two-way ANOVA was used to test the fifth null hypothesis regarding differences in the prevalence of cyberbullying offending experiences scores based on level of religiosity (higher/lower) among traditional undergraduate students attending faith-based universities (see Table 6). The null hypothesis failed to be rejected at a 95% confidence level where F(1, 280) = 0.92, p = .338, n2 = .003. The effect size was small. There was no significant difference between the prevalence of cyberbullying offending scores of traditional undergraduate students with a higher level of religiosity (M = 0.33, SD = 0.06) and the prevalence of cyberbullying offending scores of traditional undergraduate students with a lower level of religiosity (M = 0.44, SD = 0.10) who attend faith-based universities.
Results for Null Hypothesis Six
A two-way ANOVA was used to test the sixth null hypothesis regarding the interaction of prevalence of cyberbullying offending experiences scores based on biological gender (female/male) and level of religiosity (higher/lower) among traditional undergraduate students attending faith-based universities (see Table 6). The null hypothesis failed to be rejected at a 95% confidence level where F(1, 280) = 0.30, p = .586, n2 = .001. The effect size was small. There was no significant difference in the prevalence of cyberbullying offending experiences scores for traditional undergraduate female students with a higher level of religiosity (M = 0.19, SD = 0.52) and traditional undergraduate male students with a higher level of religiosity (M = 0.47, SD = 1.24) who attend faith-based universities. There was also no significant difference between the prevalence of cyberbullying offending experiences scores for traditional undergraduate female students with a lower level of religiosity (M = 0.37, SD = 0.90) and the prevalence of cyberbullying offending experiences scores for traditional undergraduate male students with a lower level of religiosity (M = 0.52, SD = 0.99) attending faith-based universities.
Table 5 Results of Two-Way Analysis of Variance: Cyberbullying Victimization
Table 6 Results of Two-Way Analysis of Variance: Cyberbullying Offending
Table of Contents
- CHAPTER ONE - INTRODUCTION
- CHAPTER TWO - LITERATURE REVIEW
- CHAPTER THREE - METHODS
- CHAPTER FOUR - FINDINGS
- CHAPTER FIVE - CONCLUSIONS
- REFERENCES
- APPENDIX A - CYBERBULLYING AND ONLINE AGGRESSION SURVEY
- APPENDIX B - DUKE UNIVERSITY RELIGION INDEX (DUREL)
- APPENDIX C - INSTITUTIONAL REVIEW BOARD PERMISSION