Regression analysis and variable selection

• Uncategorized

REGRESSION ANALYSIS AND VARIABLE SELECTION

Regression analysis is a research tool that is used in quantitativestudies to determine the relationship between a dependent variableand specific independent variables (Gray J. R., Grove S. K. &ampSutherland S., 2016). The degree of relationship between the twotypes of variables is used in developing a regression equation.Regression analysis can be used to study the relationship betweendifferent blood pressure-lowering drugs and the central aorticpressures. A study was conducted to examine the effect of differentblood pressure (BP) medications on the central aortic pressures. 2019patients in five different hospitals in Washington State wererecruited. The participants were randomly grouped into two groupsbased on the medications atenolol-thiazide andamlodipine-perindopril. The independent variables used in the studyare the two medications while the central aortic pressure is thedependent variable.

The regression analysis involved in this study is multivariate sinceit involves more than one predictor variable. When developing theregression equation, predictor variables are selected in astatistical manner. Selection is important since it helps indetermining the importance of each predictor variable. Standard andhierarchical selections are the commonly used methods. Inhierarchical regression analysis, the independent variables or blocksof independent variables are introduced into the equation in an orderdetermined by the investigator (Vogt, W. P., &amp Johnson, B. 2011).The variable sequence is dictated by the theoretical understanding ofthe relationship between the variables by the researcher. Thevariance in the dependent variable (central aortic pressure) isexplained by the independent variables (the two medications) enteredinto the regression equation. In the standard method, all theindependent variables are introduced into the equation at once(Chatterjee S. &amp Hadi A. S. 2015). When the two independentvariables used in the study overlap in accounting for the variance ofthe dependent variable, the effects of either of these independentvariables are not accounted for.

References

Gray J. R., Grove S. K. &amp Sutherland S. (2016), Burns &ampGrove`s The Practice of Nursing Research: Appraisal, Synthesis, andGeneration of Evidence, Elsevier Health Sciences

Chatterjee S. &amp Hadi A. S. (2015), Regression Analysis byExample, John Wiley &amp Sons

Vogt, W. P., &amp Johnson, B. (2011).&nbspDictionary ofstatistics &amp methodology: A nontechnical guide for the socialsciences. Thousand Oaks, Calif: SAGE.