+ Reply to Thread
Results 1 to 3 of 3

Thread: Doubt in Interpreting Regression Output in SPSS

  1. #1
    Points: 1,714, Level: 24
    Level completed: 14%, Points required for next Level: 86

    Posts
    32
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Doubt in Interpreting Regression Output in SPSS




    I am trying to understand Regression through a small example. I want to predict the sales for the year 2011, and Ive considered only two independent variables Production and Ad Expenditure for simplicity. Following is the data:

    Table 1:
    YEAR SALES(in lakhs) PRODUCTION(in lakhs) ADEXPENDITURE(GRPs)
    2008 7 11 900
    2009 8.5 12 900
    2010 9.2 12.4 1000

    Following are the Z-scores for the above:

    Table 2:
    ZSALES ZPRODUCTION ZADEXPENDITURE
    -1.09729 -1.30854 -0.86603
    0.23725 -0.11896 -0.86603
    0.86004 0.35687 0.86603

    I run Linear Regression on the data of Table1, with Sales as the dependent variable and Production and AdExpenditure as the independent variables.
    Following is the output in SPSS:


    Model Summary
    Model R R Square Adjusted R Square Std. Error of the Estimate
    1 1.000a 1.000 . .
    a. Predictors: (Constant), ADSPEND, PRODUCTION



    ANOVA
    Model Sum of Squares df Mean Square F Sig.
    1 Regression 2.527 2 1.263 . .a
    Residual .000 0 .
    Total 2.527 2
    a. Predictors: (Constant), ADSPEND, PRODUCTION
    b. Dependent Variable: SALES


    Coefficient
    Model Unstandardized Coefficients Standardized Coefficients
    B Std. Error Beta
    1 (Constant) -10.400 .000
    PRODUCTION 1.500 .000 .962
    ADSPEND .001 .000 .051


    As per the coefficients, the regression equation is:
    SALES = -10.4 + 1.5*PRODUCTION + .001*ADSPEND

    Verification:
    To verify the above linear equation, let us try to predict the sales of 2010,and 2011.
    Sales (2010) = -10.4 +1.5*12.4 + 1
    Sales (2010) = 9.8 (seems reasonable, atleast for demonstration)

    Verification (Z Score Coefficients): (Constant is always zero in this case)
    Sales (2010) = 0.962*0.35687 + 0.86603*0.051
    Sales (2010) = 0.3433 + 0.04 = 0.3833 ( no where close to the actual value 0.86 from Table 2)


    Please let me know why am I not able to get the Z Score coefficient for Sales 2010 correctly. This forum has been excellent in resolving my doubts.

    I am sorry for the untidy math here

  2. #2
    Beep
    Points: 60,883, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Discussion EnderPosting AwardCommunity AwardMaster TaggerFrequent Poster
    Dason's Avatar
    Location
    Ames, IA
    Posts
    11,034
    Thanks
    259
    Thanked 2,132 Times in 1,814 Posts

    Re: Doubt in Interpreting Regression Output in SPSS

    You only have 3 observations. You have two linearly independent predictors (ie one is not a linear transformation of the other). You are perfectly predicting the observations. This will happen anytime you only have 3 observations and try to use two predictors. If you have 4 observations and try to use 3 predictors (as long as the predictors aren't linear combinations of each other) you will be able to perfectly predict the responses.

    Take for example if you have two observations and a single predictor. It should be pretty obviously that you can perfectly fit a line through the two points. If you don't see this draw a few scatterplots using two observations.

    What it boils down to is if you have n observations and try to use n-1 predictors that aren't linear combinations of each other then you will always be able to perfectly predict the response.

  3. #3
    Points: 1,714, Level: 24
    Level completed: 14%, Points required for next Level: 86

    Posts
    32
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Doubt in Interpreting Regression Output in SPSS


    HTML Code: 
    [B]Verification (Z Score Coefficients[/B]): (Constant is always zero in this case)
    Sales (2010) = 0.962*0.35687 + 0.86603*0.051 
    Sales (2010) = 0.3433 + 0.04 = 0.3833 ( no where close to the actual value 0.86 from Table 2)
    Am I correct to verify the standardised coefficients like the above?
    If so, why am I not getting the Z score of Sales 2010, which is 0.86

+ Reply to Thread

           




Similar Threads

  1. Replies: 0
    Last Post: 03-05-2011, 03:40 AM
  2. Interpreting Logit Regression Output
    By civilsurfer in forum Regression Analysis
    Replies: 0
    Last Post: 02-28-2011, 01:08 PM
  3. Doubt on partial correlation SPSS output
    By lndtrp in forum Statistics
    Replies: 7
    Last Post: 01-27-2011, 09:31 AM
  4. Interpreting logistic regression output
    By laura_rw in forum SPSS
    Replies: 0
    Last Post: 05-15-2010, 07:43 AM
  5. SPSS output
    By bearnie0 in forum Statistics
    Replies: 1
    Last Post: 04-17-2010, 07:06 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats