Welcome to Econ 420 Applied Regression Analysis

1
Welcome to Econ 420 Applied Regression Analysis

• Study Guide
• Week Three
• Ending Tuesday, September 11
• (Note You must go over these slides and complete
  every task outlined here by noon on Wednesday
  September 14)

2
Assignment 1

• I will email you the graded assignment.

3
Answer Key to Assignment 1

• Question 1
• Use the data set on the next slide and the
  formulas on Page 8 (1-5 and 1-6) to estimated the
  coefficients ß0 and ß1 in the equation below
• W ß0 ß1 H
• Make sure to show your work.
• Do the estimated coefficients make sense to you?
• What is the meaning of the estimated
  coefficients?

4
Here is our sample data on height and weight.
5

• The estimated regression equations is
• Wi - 231 5.9 Hi
• Here are the calculations

6
Do the estimates make sense to you? What is the
meaning of the estimated coefficients?

• ß0 - 232 is the intercept which measures the
  weight when the height is zero. This does not
  have useful information in it because
• Zero height is impossible
• The intercept captures the mean effect of all
  other excluded explanatory variables (such as
  calorie intake..) on weight.,

7
Do the estimates make sense to you? What is the
meaning of the estimated coefficients?

• ß1 5.9 is slope of the line, it means,
  ignoring the effect of all other explanatory
  variables on weight, if the height increases by
  1 inch, the weight will increase by 5.9 lb.
• The sign (positive) makes sense.
• The value may not make sense a lot because when
  we exclude the other explanatory variables from
  our equation, we can not hold them constant. And
  5.9 may tend to in part capture the effect of
  excluded explanatory variables on weight.

8
2. Question 5 on Page 15

• a. ß1 is the true, theoretical value of the slope
  that expresses the relationship between X and Y.
  It is unobservable since we have a sample of
  data, and not all the data possible. ß1 is an
  estimate of ß1 that is based on the sample data
  and comes from the regression process.
• b. e is a theoretical error term that cannot be
  observed it comes from measurement error,
  factors that influence the dependent variable
  that have not been included in the model, and
  random variation. e is an actual observation of
  the error term that comes from estimating the
  regression. It is also called the residual.
• c. Y is the actual value of the dependent
  variable. Y is the fitted or predicted value of
  the independent variable.

9
3. Answer Question 8 on Page 15

• a. It means that the process finds estimates for
  the coefficients that minimize the squares of the
  errors.
• b. It indicates that OLS minimizes the sum of the
  errors squared, not just the sum of the errors.
• c. If you found the coefficient estimates by
  choosing values that minimize the sum of errors,
  you would not get the best regression line,
  because positive and negative errors would cancel
  each other out. That is why the errors must be
  squared.

10
Assignment 2

• This is an EViews assignment.
• It is due before noon on Wednesday, September 12.
• It has 20 points

11
To use EViews to estimate the coefficients of
our height weight problem

• Create and save an Excel spreadsheet
• In boxes A1, B1 and C1 type obs, w, h
  respectively
• Save your excel spreadsheet as book1
• Open EViews
• Click on File, then new, then workfile
• Choose undated
• Set the start to 1 and end to 8

12

• 5. Go to file then import then read excel
• 6. Find your excel file make sure you choose
  excel as the option under file type
• 7. EViews will give you a window in which you
  will need to type w h (make sure there is a space
  between the two)
• 8. Now on the small window click on view and
  then show and then type w h in the box
• 9. EViews will give you a window with the values
  of w and h recorded in them. Check the values
  and make sure they are correct.
• 10. Now go on the top of the page and click on
  quick then estimate equation and then in the box
  type w c h (there should be a space between w, c
  , and h.)
• We always type the dependent variable first, then
  c for constant or intercept (if you dont type
  c, EViews will estimate an equation without an
  intercept.) and then the independent variable

13

• 11. EViews will generate a table that summarizes
  the regression results.

14
Here is how your results should look

• Dependent Variable W
• Method Least Squares
• Date 09/09/07 Time 0815
• Sample 1 8
• Included observations 8
• Variable Coefficient Std. Error t-Statistic Prob.
  
• C -231.1391 92.54581 -2.497565 0.0467
• H 5.879205 1.314659 4.472039 0.0042
• R-squared 0.769223 Mean dependent
  var 181.8750
• Adjusted R-squared 0.730760 S.D. dependent
  var 32.39681
• S.E. of regression 16.81016 Akaike info
  criterion 8.694162
• Sum squared resid 1695.489 Schwarz
  criterion 8.714023
• Log likelihood -32.77665 F-statistic 19.99913
  
• Durbin-Watson stat 1.128086
  Prob(F-statistic) 0.004228

15

• Press on shift and then highlight the output and
  copy it on a word document
• Then
• Identify the values on the EViews output that you
  understand
• Identify the values on the EViews output that you
  dont understand
• Send the assignment to me before Wednesday,
  September 12 at noon as a word attachment to an
  email.
• Make sure that under the subject of the email you
  put your last name, class, asst 2.
• If it was me then the subject would have been
• Khorassani,econ420,asst2

16
On Wednesday afternoon

• Look for a new study guide on WebCt.
• Study Guide 4