THE TERM PROJECT
You will undertake a multiple regression project by applying all the tools and techniques introduced in the course. This application will apply to any question of interest to you.
As the model builder, you must use good sense and good theory in your application.
Your final report should be about 15 pages long (typed -double spaced) not including all the computer printouts (Appendices). Your report should be written using the Harvard style of referencing. I am indifferent to whether you use the APA or MLA style of writing.
The following steps will provide you with some guidelines as to how to undertake the project and how to present your final report. It is meant to supplement Chapter 11 A Regression Users Handbook of the Studemund text.
Read the chapter and these notes and determine which parts are relevant to your specific project.
Specify the dependent variable of interest. State clearly the relevance of this dependent variable to some purpose of forecasting, decision making or problem-solving
On the basis of reading, experience, theory, derivations or results from other studies, specify the explanatory variables in your model. Keep in mind the purpose of your choice of the dependent variable. Variables useful for good forecasting may differ from variables that can be manipulated for policy changes or from variables that are theoretically supposed to be causal factors.
State clearly the expected shape of the function relating each explanatory variable to the dependent variable. That is, try to specify the most reasonable functional form (linear or non-linear) and find the best measures of the variables desired in the model. Be sure to note the definitions, the units, time periods, the number of cases, and source of data.
Define the symbols Y, X1, X2
.Xm, for the linear regression model. These may be transformations (first differences, logs etc.), substitutes (proxies), dummy variables, etc.
Specify the expected sign of the regression coefficients Bj relating Y to Xj. These propositions must take into account the functional relationship and transformations, if any, described above.
Consider the possible outcomes of your analysis. Suppose your expectations are confirmed by tests on the sample model, what will you have shown? Suppose such tests and results falsify one or more of the prior suppositions, will you have anything meaningful to argue or any worthwhile information to provide relative to the purpose of your project?
Think about the underlying set of assumptions in the model. Question the validity of each of these for your model and think of situations or reasons why they might be violated
Estimation and Testing
Use a computer program and your data to obtain the results for the estimation, using the methods of least squares. Remember that you may have to estimate several models, such as models excluding one, two or a subgroup of variables, in addition to the full model. Also, include a listing of the data with labels in your report and be sure to instruct the computer to provide estimated values of the residuals for each regression. You may also want to do some plots at this stage.
Perform the standard analyses for the goodness of fit of your model, and do one-sided test of significance of the coefficients corresponding to your propositions (expectations).
Interpret these results
Complete any other useful analysis of the results
Examine the correlation coefficients among the explanatory variables for detection of multicollinearity. Examine the plots and tests of residuals for detection of autocorrelation and heteroscedasticity.
If one or more of these violations are detected a re-estimation of the model may be needed. Make clear what problems have been detected and its effects on your estimation etc.
In summary, your report should include, but not limited to the following: (Refer to page 396 of the text)
An abstract or executive summary of the project
Statement of the purpose of the project
Statement of the model(s) and its relevance to the issue
Definition of variables and source of their statistical measures detailing the units and cases being used in each regression
Hypotheses to be tested, forecast to be calculated and their relevance to the issue
Summary of results, including residual analysis and revised estimations
Conclusions based on the results and their overall bearing on the issue raised in your statement of purpose
Culled from D. Harnett and J. Murphy (1993):Statistical Analysis for Business and Economics:
________________________________________ Item Id
9 000002 Custom Writing Service $179.55 USD
Total $179.55 USD
Your order has been processed and your credit card billed.
A confirmation email detailing your purchase has been sent to [email protected]
Charges will appear on your credit card bill or bank statement under the name 2CO.COM.
Your 2Checkout.com order number is: 104490-15646381
For Questions About Your Order Contact: