Health as multicategorical independent variable
weight by pspwght.
fre rlgblg.
RECODE rlgblg (1=1)(2=0) into rlgblg_2cat.
VARIABLE LABELS rlgblg_2cat ‘Belonging or not to particular religion or denomination?’.
VALUE LABELS rlgblg_2cat 1’yes’ 0’no’.
fre rlgblg rlgblg_2cat.
fre gndr.
RECODE gndr (1=1)(2=0) into gndr_2cat.
VARIABLE LABELS gndr_2cat ‘gender=male’.
VALUE LABELS gndr_2cat 1’male’ 0’female’.
fre gndr gndr_2cat.
fre domicile.
RECODE domicil (1 2=1)(3 thru 5=0) into domicil_2cat.
VARIABLE LABELS domicil_2cat ‘domicil=big city’.
VALUE LABELS domicil_2cat 1’big city or outskirts’ 0’not big city’.
fre domicil domicil_2cat.
fre health.
LOGISTIC REGRESSION rlgblg_2cat WITH gndr_2cat domicil_2cat health
/CATEGORICAL health
/CONTRAST (health)=indicator(1).
*Note: very good is the reference category.
b3: The log odds of being religious is higher by 0,176 among those who have a good health status than among those who have a very good health status, everything else held constant.
In other words: The log odds of being religious is higher by 0,176 among women who live in small cities and have a good health than among women in small cities who have a very good health. So, the log odds of being religious is higher by 0,176 among women who live in small cities and have a good health than among people in the same categories who have a very good health.
Exp(b3): The odds of being religious among those who have a good health status are 1,192 times as high as among those who have a very good health status, everything else held constant.
Exp(b4): The odds of being religious among those who have bad health status are 3,690 times as high as among those who have a very good health status, everything else held constant. (Note: we compare them to those who have a very good health status because this is our reference category.)
Note: Bad health status is coded in the column (3) in the below table. So, that is how we know that health(3) refers to those who reported a bad health status.
As you saw above, in the syntax, we did not create manually all the dummy variables for the variable called “health”. The following table (Categorical Variables Coding) shows us how the SPSS created these dummy variables automatically.
The category that is coded everywhere with 0 will be our reference category -> very good.
In the Regression table health(1) refers to those who have a good health status, because good is coded as 1 in the column called (1).
The same way health(2) refers to those who reported a fair health status, because in the row where fair is mentioned the number 1 is in the (2) column.
health(3) in the regression table : bad health status
health(4) in the regression table: very bad health status
How can you change which category to regard as a reference category?Answer: With the subcommand “/CONTRAST (health)=indicator(1)” by changing the number in the brackets to the corresponding category number.