There may be times when you would like to run a logistic regression with no
predictor variables; in other words, just the constant (a.k.a. the
intercept). For example, one may do this when calculating deviance scores
between various models. If you try to run the logistic regression command
in SPSS without a method subcommand or a method = enter
subcommand with no variables after it, SPSS will give you an error message and
not run the logistic regression. There is a way to "trick" SPSS
into running this type of logistic regression model. First, you will need
to create a new variable that is a constant in the dataset. Next, when you
run the logistic regression, use this new (constant) variable as the independent
variable with the noconst subcommand. In effect, you are simply
substituting the constant that you create for the one that would normally be
included in the model. (Please note that this trick does not work with the
regression command. According to SPSS technical support, the
regression command cannot be run without predictors; in other words, you
cannot get an intercept only model. If you want an intercept only model,
you will need to use the glm command.)
For example, let’s use the /spss/faq/hsb2.sav
dataset. First, we will create the constant
variable. Next, we will run the logistic regression using female as the dependent
variable (we understand that this is an unusual choice for a dependent
variable, but we just needed a dichotomous variable for the example).
compute constant = 1. execute.
logistic regression var = female
/method = enter constant
/noconst.
Case Processing Summary
Unweighted Cases(a)
N
Percent
Selected Cases
Included in Analysis
200
100.0
Missing Cases
0
.0
Total
200
100.0
Unselected Cases
0
.0
Total
200
100.0
a If weight is in effect, see classification table for the total number of cases.
Dependent Variable Encoding
Original Value
Internal Value
male
0
female
1
Classification Table(a,b,c)
Predicted
FEMALE
Percentage Correct
Observed
male
female
Step 0
FEMALE
male
0
91
.0
female
0
109
100.0
Overall Percentage
54.5
a No terms in the model.
b Initial Log-likelihood Function: -2 Log Likelihood = 277.259
c The cut value is .500
Variables not in the Equation
Score
df
Sig.
Step 0
Variables
CONSTANT
1.620
1
.203
Overall Statistics
1.620
1
.203
Omnibus Tests of Model Coefficients
Chi-square
df
Sig.
Step 1
Step
1.622
1
.203
Block
1.622
1
.203
Model
1.622
1
.203
Model Summary
Step
-2 Log likelihood
Cox & Snell R Square
Nagelkerke R Square
1
275.637
.008
.011
Classification Table(a)
Predicted
FEMALE
Percentage Correct
Observed
male
female
Step 1
FEMALE
male
0
91
.0
female
0
109
100.0
Overall Percentage
54.5
a The cut value is .500
Variables in the Equation
B
S.E.
Wald
df
Sig.
Exp(B)
Step 1(a)
CONSTANT
.180
.142
1.616
1
.204
1.198
a Variable(s) entered on step 1: CONSTANT.