Originally Published MX January/February
2003
The Terminology of Statistical Analysis
|
Return
to Article: |
Multiple regression is a method of explaining the linear relationship between several independent, or predictor, variables and a dependent, or criterion, variable. Multiple regression enables a researcher to seek a trustworthy answer to the question, What is the best predictor of ______? The statistical analytical technique determines the linear relationship between the values that change, and then finds an equation that satisfies such a relationship.
R-square is an indicator of how well the model fits the data, and is a value expressed as a decimal term between 0 and 1. An R-square of 0.4 means that 40% of the
original variability has been explained, and the re-maining 60% cannot be explained by these variables in the equation.
Coefficients are the multipliers in front of the independent variables in the equation. The higher their value, the more impact on the dependent variable they are trying to predict (in the case presented here, market capitalization). Those with especially high values are referred to as
driving coefficients because they contribute disproportionately to the value in an equation of the independent variables with which they are associated. This does not imply, however, that they are meaningful or statistically significant (see statistical significance, below).
T statistics for each coefficient explain the statistical significance that can be attributed to its independent variable in the regression equation, that is, the confidence one can have that the variable’s coefficient is truly meaningful and likely.
Statistical significance is the degree of assurance that the value assigned an independent variable in a regression equation is likely and meaningful. It does not necessarily mean that the statistically significant factor is important, only that it is very probable.
Copyright ©2003 MX
