Nothing Special   »   [go: up one dir, main page]
Scribd Logo
Upload

Welcome to Scribd!

  • 51 views

    CC 2 Multiple Regression 0

    Uploaded by

    Edward Day
    This document provides instructions for using Eviews software to estimate a multiple regression model examining factors that influence birthweight. It describes estimating a regression of birthweight on family income, smoking during pregnancy, and birth order. The output is interpreted, including coefficients, test statistics, fit measures, and conducting an F-test on a subset of variables. Graphs of actual, fitted, and residual values are also examined.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as DOCX, PDF, TXT or read online from Scribd

    CC 2 Multiple Regression 0

    Uploaded by

    Edward Day
    51 views3 pages
    This document provides instructions for using Eviews software to estimate a multiple regression model examining factors that influence birthweight. It describes estimating a regression of birthweight on family income, smoking during pregnancy, and birth order. The output is interpreted, including coefficients, test statistics, fit measures, and conducting an F-test on a subset of variables. Graphs of actual, fitted, and residual values are also examined.

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document provides instructions for using Eviews software to estimate a multiple regression model examining factors that influence birthweight. It describes estimating a regression of birthweight on family income, smoking during pregnancy, and birth order. The output is interpreted, including coefficients, test statistics, fit measures, and conducting an F-test on a subset of variables. Graphs of actual, fitted, and residual values are also examined.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    51 views3 pages

    CC 2 Multiple Regression 0

    Uploaded by

    Edward Day
    This document provides instructions for using Eviews software to estimate a multiple regression model examining factors that influence birthweight. It describes estimating a regression of birthweight on family income, smoking during pregnancy, and birth order. The output is interpreted, including coefficients, test statistics, fit measures, and conducting an F-test on a subset of variables. Graphs of actual, fitted, and residual values are also examined.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    You are on page 1of 3

    Computer Class 2: Multiple Regression

    Quantitative Economics III (L12306)


    Computer Class 2: Multiple Regression in Eviews 7.0
    Today we will . . .
    1. Estimate a multiple regression equation.
    2. Interpret statistical output (coefficients, t statistics, p-values, R2 from the estimated model)
    3. Conduct F tests, plot the fitted values and residuals and interpret results

    At the start
    Log on to the University Network. Go to WebCT and get the Word Document CC 2 Multiple Regression and save
    this to your Z Drive as CC2 Report. While you are there upload a copy of the statistical tables as we will use these
    later. To Launch Eviews 7.0 go to the Start menu and follow links from All Programs/UoN Software/_School
    Economics/Eviews 7.0. We are going to use the workfile that we saved in your Z: Drive at the end of last week so
    go to the File menu/Open/Eviews Workfile to obtain it. A copy is also available on ‘Public on
    ‘Apps2\Apps2\Sys’ (Y:)/Eviews Data/L12306/bwght1.wf1 When the workfile has been loaded it will appear
    on the screen. Click and drag it to expand it across the screen beneath the command line.

    The Model
    Health specialists suggest that birthweight (a measure of a baby’s health) is determined by a wide range of
    biological, social and economic factors. Among these are the income of the family in to which the baby is born
    (faminc), the birth order of the child (order) and the smoking habits of the mother during pregnancy (cigs).
    Today we will estimate a regression model using these variables. We will suspend reality for the moment and
    assume that these are the only variables that affect birthweight. Clearly, this is untrue (so we are likely to violating
    MLR4 of the CLM, the other variables in the workfile are other potentially important factors that you could explore
    in your own time). For simplicity we’ll formulate the model of birthweight as:

    bwght i =β0 + β 1 faminc + β 2 cigs+ β 3 order + μi

    What signs would you expect on the coefficients?

    Data Description
    Click on Details+/- to get a description of each of the variables in the workfile. Click on each of the four objects
    (birthweight, cigs faminc, order) in turn. Graph the variables and familiarise yourself with what the data
    actually mean, making sure you understand the units each is measured in. Select all four at the same time (with the
    control key) and click View/Open Selected/One Window/Open Group. Having got the four series together go to
    View/Descriptive statistics/Common sample to see summary statistics. Name this group Key_variables
    with the description Variables used in Equation 1,. Copy the table and paste it below.

    Insert Summary Statistics here

    Interpret the mean value of these variables.

    1
    Computer Class 2: Multiple Regression

    Estimating Regression Equations


    Close the Group window containing the key variables and go to Object/New Object/Equation and specify the
    model

    bwght c cigs faminc order

    Saving Regression Output and Exporting to Word


    Click Okay and the model is estimated with results displayed in the table. Before we interpret the results we will
    make an object of the regression equation. When we estimate another equation the new results will populate this
    object (i.e. the table is updated with results from the latest regression). With the results open as the active window
    click Name and type

    Latest_Equation

    We will also create a permanent record of the results from this equation, that we can return to later, so with the
    results open as the active window, click Name and call it Equation_1 with the description:

    Regression of bwght on c faminc parity cigs.

    To get a description of the model in equation form rather than as a table double click on the Latest _Equation
    Object, click on View and select Representation. Also note that an alternative way of exporting results in to a Word
    document is simply to use Select All and Copy with the results in the active window..

    Insert regression table here

    Note that in the list of objects in the workfile these two objects have the equation icon. We could have saved
    the results as a table by clicking on Freeze and then naming it. This produces a fixed record (graph, table etc) with
    the icon. Objects with the icon are probably more useful as they allow you to change the view (output,
    graph etc) whereas icons cannot be viewed as anything else.

    Interpreting Regression Results


    There are over 35 pieces of information (some of which you will not have come across before). Take some time to
    interpret what has been found. Answer the following:

    (i) The estimation is based on the information of how many babies?

    (ii) Explain in words the literal interpretation of each coefficient that has been estimated? Do the signs
    concur with your a priori expectations?

    (iii) Test the null hypothesis of ‘no relationship’ for each coefficient using the t statistics

    (iv) What do the p-values of the regression coefficients tell us?

    (v) Interpret the standard error of the regression, R 2 and the F statistic for the significance of the
    regression from this model.
    2
    Computer Class 2: Multiple Regression

    Conducting F tests
    We have already used the F test to calculate the ‘the significance of the regression’, which in our case was a test of
    the joint null hypothesis .that H 0 : β1 =β2 =β3 =0 against the alternative that at least one of the coefficients is
    different from zero. This is such a common use of the F test that it is reported automatically as in the regression
    output here. However, the F test is a very flexible test and can be used in all manner of situations.

    Let’s say we want to test the significance of a subset of the variables in a model, for example the income and
    cigarette variables i.e. H 0 : β1 =β2 =0, (Given that they are individually significant, they will also be significant as a
    group, but we will test them anyway). We could test this null by estimating a (so-called restricted) model in which
    the two variables are excluded (by restricting β 1∧β 2 to zero) and comparing the model fit (as given by Sum of
    Squared Residuals) with the original unrestricted model. We could form the F statistic of this null hypothesis,
    which is distributed as an F variate on q and n-k-1 degrees of freedom if the null is true, i.e.

    (SSR r −SSR u )/q


    F= ~ F q (0 . 05 ) if H 0 is true
    SSR u /n−k −1 n−k−1

    Large values of the test statistic is evidence against the null (implying that the null is false and the variables are
    jointly significant). Rather than calculate the F statistic ‘by hand’ using the SSR from the two models, note that
    Eviews computes the F Test automatically. With the Latest_Equation results window active, go to View,
    Coefficient Diagnostics/Wald test– coefficient restrictions and type:

    C(2)=0,C(2)=0
    Interpret the result from this F test. To make sure that understand what Eviews has done, try calculating the F
    statistic manually and compare the result with the F distribution at 5% significance. You should get an identical F
    Statistic to that obtained using the automated procedure.

    Obtaining other Output in Eviews


    As you will see from navigating through the menus, there is a wealth of statistical output that is available to you.
    There are also a number of other useful things to look at in the View menu. For example:

    View, Coefficient Diagnostics/Confidence Intervals gives a table of the confidence intervals of the estimate.

    Interpret the result.

    View, Coefficient Diagnostics/Actual, Fitted and Residual/Actual Fited Residual graph

    What does this tell you?

    Saving the data and Exit


    Finally, Save the workfile to your Z:Drive as bwght2.wk1 and then exit from Eviews 7.0. Don’t forget to log out of
    the network.

    ______________________

    You might also like