Rates of Bullying in the USA - Assignment Example

Rates of Bullying in the USA Introduction Bullying is among the most important behavioral problems with which many school administrators must deal. Because it can be found at all levels of education and must be controlled in order to serve a school’s socializing function, as well as in order to maintain discipline and promote a healthy learning environment, it is important to recognize the factors that are associated with the rates of prevalence for school-based bullying. This brief paper will present a research design to study a number of factors that are associated with rates of bullying in order to suggest an experimental statistical research project that can be used to outline the problem and suggest possible solutions. The relevant data for the research design will be taken from the United States Statistical Abstracts -- specifically details regarding incidents of crime and safety concerns in American public schools. By presenting the outlines of a statistical design relevant to this data, including an initial tabular presentation of the descriptive statistics regarding the data, it is expected that this brief paper will illustrate both an understanding of basic statistical processes and a possible experimental approach to understanding the important social problem of bullying. Variables The dependent variable in the proposed research will consist of the rate of bullying evident in a variety of schools having different characteristics, expressed as a percentage. As such the dependent variable will be a ratio-type variable, having the ability to be expressed as a measurable number with fine distinctions possible as to quantity. The independent variables to be utilized for this study will be drawn from data provided in the Statistical Abstracts, including age-level of students (expressed as an ordinal variable measured as elementary, middle, of high school), enrollment of school (expressed as ordinal breakdowns determining the size of schools), and minority enrollment within schools (again expressed as an ordinal breakdown). It is expected that these different independent variables will each have an impact on the rate of bullying, as they determine the developmental age of students, as well as the size and supposed social stresses and intercommunity cohesion of the various school types. The proposed relationships of the variables will be as defined in the following section. Research Problem and Hypothesis The basic research problem will be to discover how the age and population characteristics of a given school are associated with incident rates of bullying. It is expected that the older the student population, the more likely it will be that bullying will be prevalent, largely because the students will be more likely to be willing to use coercion and violence against their peers as they grow into their own independence. This leads to the following hypothesis. H1. The older the student population, the higher the rate of bullying that will exist. Similarly it is expected that the larger a school is the more likely bullying will be prevalent, because (1) the school community will be more difficult to govern, and (2) the larger number of students will likely include more of the personality type that is prone to violence and bullying. This results in the second hypothesis. H2. The larger the school, the higher the rate of bullying that will exist. Finally, it is expected that the higher the rate of minority students within a given school, the more likely bullying will be prevalent. It is expected that two factors will work toward this end. First, the prevalence of more diverse communities will likely result in more social cohesion problems at this developmental age. Second, it is expected that minority students will be more likely to be involved in bullying, either as its victims, or as perpetrators, as they seek to express their own independence. This leads to the third hypothesis. H3. The larger the minority enrollment, the higher the rate of bullying that will exist. Research Design In order to test the hypotheses an appropriate statistical measure must be chosen. While the student t-test has often been used in the evaluation of such data, some findings show that is inappropriately applied to data that does not include ratio level independent variables Romano et al., 2006). Therefore the Mann-Whitney U test will used as it has been shown to be a robust predictor for data that includes ordinal level independent variables (Lehman, 1999). This statistical measure also has the added advantage of being calculable using common programs such as MS Excel. Table 1 presents the descriptive findings for a preliminary analysis of the data. While it seems likely from a review of this finding that none but the H2 hypothesis will prove significant, the data does suggest that there are clearly differences in student population types that are associated with different rates and prevalence of bullying. For the data as presented, it seems likely that a statistical treatment of H2, at least, would yield a serviceable conclusion relevant to the proposed direction of the relationship. Table 1. Rates of Bullying by School Type School characteristics Student Bullying All public schools 25.3 Level Primary 20.5 Middle 43.5 High School 21.7 Combined 24.9 Enrollment >300 18.7 300-499 20.8 500-999 30.6 1,000 or more 33.2 Percent minority enrollment > 5 percent 25.6 5 to 20 percent 24.9 20 to 50 percent 22.1 50 percent or more 27.6 Conclusion The research design presented needs refinement. It may be the case, for example, that above and below certain thresholds for both developmental age and minority enrollments there are more likely to be problems. A more thorough research analysis than can be provided here can help determine this. If the proposed study were conducted, it is expected that H2 would be supported, while suggestive findings regarding H1 and H3 would also be defined for further analysis. References Lehmann, E. (1999). Elements of Large Sample Theory. Springer. Romano, J., et al. (2006). Appropriate statistics for ordinal level data: Should we really be using t-test and Cohen’s d for evaluating group differences on the NSSE and other surveys? Paper presented at the annual meeting of the Florida Association of Institutional Research, February 1 -3, 2006, Cocoa Beach, Florida United States Statistical Abstracts. (2007-2008). [online] http://www.census.gov/compendia/statab/cats/education/school_crime_and_safety.html [Accessed 3 August 2011]. Read More