Acknowledgments
Download 3.57 Mb.
|
|
Once again, before computing  and  it is beneficial to plot the scatter plot to determine if the correlation between the variables is approximately linear. If the data appears linear, statistical computer packages can aid in the calculations [2]. In this project, Microsoft Excel was used.
VARIABLE INTERACTION Many times there are occurrences when a number of variables are proposed to be connected to others. In other words, the variables are not always independent of each other [4]. In the context of this project, one’s gender may influence the sport played. For instance, if you are a male you cannot play softball, volleyball, etc. Thus, perhaps the two variables are not independent. One way to approximate the independence of two variables is to find the correlation between the two. While correlation does not imply causation, plotting the observed values, finding a regression model, and examining the correlation coefficient can provide more insight into the possibility of interaction between the variables. In Section 2.2.2, the differences in multiple regression models with and without interacting variables can be seen. RESIDUAL ANALYSIS When determining whether or not the regression model chosen is an adequate model to fit the data, it is beneficial to construct a diagnostic plot. One specific diagnostic plot places the predicted value for  ,  on the vertical axis and the actual value,on the horizontal axis. Once the plot is constructed, if the points closely fit a line with slope of positive one and the line goes through the origin, then the regression model for the data gives an accurate prediction of the values actually observed.
Download 3.57 Mb. Share with your friends:
|