The Effect of Video Games on High School Students' Grades (Dissertation Sample)

The Effect of Video Games on High School Students’ Grades
Chapter 3: Research Methods
Introduction
This chapter gives a detailed description of the design used in the study, the research guiding questions, the data collection methods, data analysis and the ethical considerations in surveying high school students. Additionally, the characteristics of the sample as well as the justification for all the choices made will be presented.
Purpose of the Study
The purpose of the study is to determine whether there is data to support the premise that excessive video game playing can have a deleterious effect on high school grades. Excessive video game playing by many high school students takes most of their recreational time. A significant number of these students are in fact unable to control their urge to indulge in a video game whenever they have a minute to spare (Cohen, 1989; p. 116). Consequently, this has a toll on their social and academic lives. Excessive video games playing coupled with personal, mental and motivational factors do affect academic achievement in high schools (Dodge, 2003, p. 89). This study seeks to explore the potential relationship between video gaming and high school grades measured by GPA. To achieve this, a survey of high school students will be undertaken.
Research Null Hypothesis
Based upon the aforementioned purpose of study, one major research null hypothesis may be drawn:
H0: There are no statistically significant differences in students’ high school GPAs when they engage in excessive video game playing and when they do not.
H1: There are statistically significant differences in students’ high school GPAs when they engage in excessive video game playing and when they do not.
Research Design
The researcher will use quantitative analysis together with linear correlational analysis. Correlational research method is fitting bearing in mind that this study seeks to explore the connection between the affirmed variables and there is no intention to infer that there is any causal influence working in either direction. Moreover, this approach is appropriate because the respondents are not being classified into any particular group where their responses could be manipulated but are reporting about their perception on the possible effects of video game playing on their academic performance. Finally, linear regression is appropriate for this study because linear regression analysis allows for the appraisal of important elements of statistical relationships.
Multiple regression and linear regression do not have much difference but the application of an equation draws a line for measuring; y = b1*x + c, where c is a constant, y is an estimated dependent variable, b is a regression coefficient, and x is an independent variable (Fidell, & Tabachnick 2006, p. 45-89). In linear regression unlike multiple linear regressions, it is not important to assess the lack of multi-collinearity due to the predictor variable.
Beta coefficients are utilized to determine the level of effect of the prediction of the independent variable as opposed to the use of t test in the determination of the predictor’s importance. In the case of a significant predictor and all units of predictor, dependent variables either increase or decrease with the range of unstandardized beta coefficients (Fidell, & Tabachnick 2006, p. 48-53).
Assumptions concerning and involving linear regression require linearity and homoscedasticity to be assessed. Linearity assumes a direct link between a predictor and a criterion variable therefore resulting in a straight line while homoscedasticity assumes that the distribution of scores takes place along specific points on the line. Scatter plots are used to examine homoscedasticity and linearity (Stevens, 2009, p. 23-25).
Participants
The participants in this research will be a randomly selected sample of male and female students drawn from high schools scattered across the United States. The population will be diverse in ethnicity as well as language and almost all participants will be aged between 13 and 18. The participants will be selected on the following basis: a) they should be high school students presumed to have some experience in video gaming; b) they should be available c) they should present the population that is targeted by the study and d) they should have the intellectual capacity to understand and respond adequately to the contents of the questionnaire.
Sample Size
It is important to use an adequate sample size relative to the goals of a study. The larger the sample size the greater the probability of finding significance in linear regression. Determining an adequate sample size increases the probability that a statistical investigation will rebuff the null hypothesis or increase the capacity of a statistical test to discover an effect (Tabachnick & Fidell, 2006, p.13-22). However cost considerations dictate that as small a sample size as possible be used as long as precision is not significantly affected. A power analysis has been conducted to define a sample size by setting alpha = 0.05, 0.8 power, and medium size effect (f2 = 0.15) (Erdelder, Faul, Lang & Buchner, 2008, p. 290-311). Based on the analysis, a sample size of at least 55 is required to detect a significant model.
In most cases, an alpha value below the standard 0.05 is always rejected for null hypotheses, but in special cases, type I error occurs if the null hypothesis is true. Considering the research questions, evidence is being sought to reject the null hypotheses and this gives a margin of error. Jacob Cohen (1989) argued that all null hypotheses are false in their double-tailed forms. Investigation of an issue from a sample of a particular size is aimed at finding evidence to support or reject a hypothesis (Richard, 1984, p. 114-116).
A large enough sample will provide reliable effect in the process of investigation (Stevens, 2009; Tabachnick & Fidell, 2007, p. 76).
When using the power analysis, it must be noted that there is always a possibility of an error (Trochim & William, 2006, p. 200). The hypotheses we strive to prove or reject have a potential of being the actual finding of the research in the case of type I error and type II error. A true null hypothesis can be rejected in the case of type I error, while a false null hypothesis can be accepted in type II error. In order to reduce chances of error with the use of power analysis for linear regressions (Vance, 2011, p. 83), the sample size of 55 participants is found to be sufficient for this study. However, a sample size of 70 will be chosen to allow for drop-outs and participants who do not record all data.
Data Analysis
Data collected from the sample through a survey that will also include demographic items, items inquiring about student’s technology use, and items that will be included for analyses are organized and fed to SPSS software version 18.0 for Windows (Mac operating systems are avoided for simplicity reasons). As a form of the mixed method of data collection, qualitative and quantitative methods are merged to give descriptive statistics of the data collected (Abelson, 1995). In regard to the research sample size and variables for consideration, descriptive statistics are useful, for they outline variables, specify constants and also differentiate consistent elements with random ones. The population of students reviewed can be monitored or defined to fit in a specific margin and this is classified under consistent elements. The curriculum is a major appraisal mechanism defining grading for the vast number of students and this is a constant that descriptive statistics outlines clearly (Adèr & Mellenbergh, 2008, p. 17). Whether games affect grades of high school students or not, their mental abilities, time spent playing games, time spent studying and efforts drawn to achieve better grades apply as a unit of varying elements whose definitive characteristic depends on the nature and type of an individual. Regarding the use of descriptive statistics, an in-depth analytical approach to determining frequencies of variables plays an important role for data analysis.
As a critical requirement for research data analysis, calculations are to be developed considering the nature and amount of data. When considering the size of the sample, a small one requires a model for calculating different elements and variables. Considering that there are a number of factors influencing grades for high school students, only those related to gaming are relevant for this research. To achieve the right results and to present a reliable data analysis, frequencies of time, influence, character and a need to play games are calculated alongside the respective percentage of the sample for nominal (categorical/dichotomous) data.
Continuous data are calculated following means/standard deviations; this is effective and important in determining ratios and intervals of occurring. In data analysis, these calculations play an important role, since an appraisal for occurrence in one group or setting can be compared with another one to draw a line (ASTM International; 2002, p. 35-37). An interval of occurrence is important since dependent variables are determined and identified for a better and concise analysis. Through nominal and continuous data calculations, research findings are traced back to physical numbers and descriptive statistics clarifies conflicting variables (Chow, 1996; p. 77).
Research Questions
Research Question 1
Controlling for parental income level, does the ratio of hours spent playing video games to hours spent studying predict high school GPA?
Research Question 2
Controlling for parental income level, do hours spent playing video games predict hours spent studying?
Research Question 3
Cons...