Multiple regression is a parametric technique that determines if the behavior of a particular variable can be predicted by other two or more independent variables. Tentatively, this is an extension of the linear bivariate regression which assesses the relationship between two variables (Vogt, 2006). When integrated in social research, multiple regression can be used to make predictions on different variables based on theoretical predictors associated with that variable. A key strength of multiple regression is that, it is able to determine the predictor variables that significantly affect the dependent variable, as well as, the direction and magnitude of the effect (Fox, 2015).
A study by Tahir and Naqvi (2006) assessed the factor that affected the performance of students using a case study of private colleges in Bangladesh. A sample of 300 students were subjected to the study questionnaires that focused on collecting information on personal attendance of classes, age, strictness, attitudes, hours of study after college schedules, family income, and previous academic performance of various subjects. In this case, the authors used multiple regression to assess the independent factors that best explained the student performance (Tahir & Naqvi, 2006). Notably, out of the many listed factors above, there are chances that some of them did not have a statistical significance in affecting the performance of the student – and regression was the considerable technique to achieve this goal.
Delegate your assignment to our experts and they will do the rest.
Multiple regression was the suitable choice in this case comparing with other techniques that assess linear association of variables such as bivariate regression, chi-square, and correlation. For instance, this technique has a statistical power to determine which among the theorized factors significantly explained changes in the student’s performance – which is the dependent variable in this case (Vogt, 2006). With any other statistical technique, it is easy to find an association, though it will be hard to determine which ones are significant. The statistical significance is important because it shows a causality impact. In other words, it tests cause-effect relationship for linearly related variables. With this in mind, a technique such as correlation is ruled out. Further, the fact that the variables are measured in ratio and ordinal scale, it then tells that chi-square is not fit for this study since it measures association between categorical variables (Fox, 2015). Nevertheless, the number of predictors are more than one – and this makes bivariate regression not suitable – leaving the multiple regression the only technique suitable for this study.
The authors displayed the study findings. Noticeably, the authors included tables that summarized the potential relationships between the students’ performance and the predictor variables including class attendance, family income, hours committed in studies after college normal schedule, and mother’s age, and education too (Tahir & Naqvi, 2006). This table gave the possible expected relationship and a justification – and this prepares the reader on the type of results to expect too. The table again is based on theoretical grounds meaning that, by its own, it is able to communicate the various theoretical accounts that explain how the variables are related. However, this table is not sufficient. The authors provided the data findings table which presented the actual study findings making it easy to compare with the expectations table. Noticeably, the findings table provided the regression output model with the important components such as the R-squared (that assesses the variation degree of the model explained by the predictors), the F-statistic (used to determine if the model is significant), and the coefficients with their respective t-statistics.
All these tables clearly show that the results stand alone in telling how the predictor variables are associated with the dependent variable. However, it is hard to tell which ones are significant from a mere look of the findings since the authors left out the p-values for the coefficients of the predictors, which are essential in testing the hypothesis whether the coefficients are different from zero (Warner, 2008). With this aspect left, it requires the reader to read through the work to establish which predictors were significant.
References
Tahir, S., & Naqvi, S. R. (2006). Factors affecting students’ performance. Bangladesh e-journal of sociology , 3 (1), 2.
Fox, J. (2015). Applied regression analysis and generalized linear models . Sage Publications.
Warner, R. M. (2008). Applied statistics: From bivariate through multivariate techniques . Sage.
Vogt, W. P. (2006). Quantitative research methods for professionals in education and other fields. Columbus, OH: Allyn & Bacon .
