UCLA Academic Technology Services HomeServicesClassesContactJobs
Stat Computing > Statistica > Textbook Examples > Computer-Aided Multivariate Analysis, 4th
 Search

Statistica Textbook Examples
Computer-Aided Multivariate Analysis, Afifi, Clark and May
Chapter 12: Logistic regression

Page 283 The coefficients at the top of the page

File
  Open - open the depression data set
Statistics
  Advanced linear/nonlinear models
    Nonlinear estimation
      Quick logit estimation
        OK
          Variables - select cases as the dependent variable and age 
          and income as the independent variables
            OK
              OK
RESULTS
  OK
    "Quick" tab
      Summary: Parameter estimates
Model: Logistic regression (logit) N of 0's:244 1's:50 (depress.sta) Dep. var: CASES Loss: Max likelihood Final loss: 127.01304480 Chi-square(2)=14.098 p=.00087
  Const.B0 AGE INCOME
Estimate 0.0280 -0.0202 -0.0413
Odds ratio (unit ch) 1.0284 0.9800 0.9595
Odds ratio (range)   0.2389 0.0739

Page 283 Figure 12.1  Logistic function for the depression data set

NOTE:  We were unable to reproduce this graph.

Page 285 Table 12.1  Classification of individuals by depression level and sex

Statistics
  Basic statistics
    Tables and banners
      "Classification" tab
        Specify tables (select variables) - select sex for list 1 and cases for list 2
          OK
            OK
Summary Frequency Table
Table: SEX(2) x CASES(2)
  SEX CASES CASES Row
  male 101 10 111
  female 143 40 183
  All Grps 244 50 294

Page 286 Odds ratios and coefficients

Insert
  Add variables - add one variable after sex called sex1.  In the box at the bottom, type = sex - 1
    OK
Statistics
  Advanced linear/nonlinear models
    Nonlinear estimation
      Quick logit regression
        OK
          Variables - select cases as the dependent variable and sex1 as the independent variable
            OK
              OK 
RESULTS
  OK
    "Quick" tab
      Summary: Parameter estimates
Model: Logistic regression (logit) N of 0's:244 1's:50 (depress.sta) Dep. var: CASES Loss: Max likelihood Final loss: 129.69883489 Chi-square(1)=8.7268 p=.00314
  Const.B0 sex1
Estimate -2.31254 1.038570
Odds ratio (unit ch) 0.09901 2.825175
Odds ratio (range)   2.825175

Page 287 Table of coefficients and standard errors

NOTE:  When selecting the response codes, it is important that you enter 1 0 because, as mentioned on page 282, the coding for the depressed and nondepressed groups have been reversed.  If you enter the response codes as 0 1 or if you click on the "Select all" button, the coding of the dependent variable (cases) will not be reversed and the signs of the coefficients from the logistic regression will be opposite to those shown in the text.

Statistics
  Advanced linear/nonlinear models
    Generalized linear/nonlinear models
      "Quick" tab
        Logit model
          OK
            Variables - select cases as the dependent variable and age and income 
            as continuous predictors
              OK
                Response codes - 1 0
                  OK
RESULTS
  Estimates
CASES - Parameter estimates (depress.sta)
Distribution : BINOMIAL Link function: LOGIT
  Level of Effect Column Estimate Standard Error Wald Stat. p
Interc   1 0.028 0.487 0.003 0.954
AGE   2 -0.020 0.009 5.139 0.023
INCOME   3 -0.041 0.014 8.651 0.003
Scale     1.000 0.000    

Pages 288-305 have been skipped for now.

Page 307 Figure 12.5  Percentage of individuals correctly classified by logistic regression

NOTE:  We were unable to reproduce this graph.

Page 307 Figure 12.6  ROC curve from logistic regression for the depression data set

NOTE:  We were unable to reproduce this graph.

How to cite this page

Report an error on this page

UCLA Researchers are invited to our Statistical Consulting Services
We recommend others to our list of Other Resources for Statistical Computing Help
These pages are Copyrighted (c) by UCLA Academic Technology Services

The content of this web site should not be construed as an endorsement of any particular web site, book, or software product by the University of California.