Quasi-Experimental and Statistical Tests

Quasi-Experimental and Statistical Tests

            A quasi-experiment is an experimental study that is used in estimating the impacts caused by specific interventions on the targeted population without assigning them treatments randomly. In this study, both the control groups and the treatments are not comparable. Thus, quasi-experiments are usually subject to internal validity (Rockers et al., 2015). The quasi-experimental design shows how the independent variable is being manipulated without assigning the participants to random treatments. An example of a good quasi-experiment is assigning students to a scholarship program randomly in control of the other variables (Rockers et al., 2015). Quasi-experiments contain procedures, outcome measures, and experimental units without random designating of the treatments. In this research, the independent variable is the rate of dropout in doctoral programs.

The price of dropouts in many colleges across the United States is as a result of doctoral scholar attrition. Doctoral scholar attrition has negative impacts on both the Universities and students (Devos et al., 2017). In this research, quasi-experiment is used to investigate how the rate of school dropout is manipulated by factors like; unavailability of finances, gender, ethnicity difference, location of the school, the school ranking, and marital status. In this research, the statement of the problem is the existence of the doctoral scholar attrition, which has negatively impacted both the Universities and students (Devos et al., 2017). For instance, the doctoral scholar attrition has contributed to the increased rate of scholar dropouts in many colleges across the United States.

Also, the Universities lack enough number of students hence, receive fewer tuition fees, which make them run without enough finances. In this scenario, the quasi-experiment can be used to determine what factors contribute to school dropout by assigning different treatments to the students (Rockers et al., 2015). The obtained data will therefore, be statistically analyzed by the use of SPSS software. Quasi-experiments are quantitative research studies. The quantitative data obtained from the research provides, objective, quantifiable, and is easier to interpret the final results.

Statistical Tests Used To Validate the Research Questions in Quasi-Experiments

Several statistical tests can be applied in the validation of the research questions. For example, the first research question is on the reasons that made the undergraduate scholars withdraw from the doctoral degree programs. To test this, I will use a multiple regression analysis (Huitema, 2011).  This test is usually used if several factors need to be tested in connection to an outcome. Multiple regression analysis is often used when the impact of several factors is being observed. In this research, several factors led many undergraduate scholars to withdraw from their doctoral programs across the United States. These factors can only be tested using multiple regression analysis tests.

The second research question is on the examples of significant machine learning techniques that can be applied to help predict doctoral students’ retention. In this case, I will use a regression test. As per Huitema (2011), regression is a statistical test that is used to find out whether a particular variable is a predictor of another. All the available machine learning techniques are therefore analyzed to show how they predict the retention of the doctoral students. The third research question is whether there is any gender variation in dropout. In this case, I will use ANOVA to test gender variations in dropouts (Huitema, 2011). This test is usually used when comparing two, three, or more variables. I this case, I will examine the ratio of the males to females who have dropped out of their scholar programs.

The rate obtained will help me know who is falling out of school more than the other. When determining the role of cumulative GPA in drop out scenarios, I will use a regression test method to determine what role GPA will play in slide-out situations (Huitema, 2011). To examine if financial assistance plays a significant role in dropout, regression test can be used. In the sixth research questions, I will use ANOVA to show the difference between the ethnic groups and their comparison to the rate of dropout (Huitema, 2011). In the seventh research question on the factors associated with institutions, I will use multiple regression analysis to determine those factors. This test method is used when several variables are tested concerning the final results.

In the eighth research question on the other factors associated with dropouts apart from kids and marriage, I will use multiple regression analysis test (Huitema, 2011). The reason is that there are several factors which are channeled to one outcome, and this can only test by the use of various regression analysis test method. The last research question is on which machine learning model will give the best prediction in terms of accuracy. In this case, I will use regression as the statistical test method (Huitema, 2011). This is because a regression test is used to find out whether a particular variable is a predictor of another variable. The best machine learning model that will give the best prediction will be shown from the list of several models. This can be done by calculating the frequencies of the different models that are proposed by the various scholars. The one with the highest rate is, therefore determined as the best model for prediction in terms of accuracy.