*This is a basic regression simulation program for SPSS. 
*First we create a working file with 500 observations.
MATRIX. 
COMPUTE CONST=MAKE(500,1,1). 
SAVE CONST /OUTFILE=*/VARIABLES=CONST. 
END MATRIX. 
EXECUTE. 
*Suppose we want to study designs when r square is 10% in simple sample.
* We tell SPSS to create 2 independent variables with normal(0,1) distributions.
COMPUTE Z1 = RV.NORMAL(0,1). 
COMPUTE Z2 = RV.NORMAL(0,1). 
* We create a new variable that is 10% of one variable.
COMPUTE X1=Z1+3.
COMPUTE E=3*Z2.
COMPUTE Y1= X1+E.
EXECUTE.

GRAPH
  /HISTOGRAM(NORMAL)=x1 .
REGRESSION
  /DESCRIPTIVES MEAN STDDEV CORR SIG N
  /MISSING LISTWISE
  /STATISTICS COEFF OUTS R ANOVA
  /CRITERIA=PIN(.05) POUT(.10)
  /NOORIGIN
  /DEPENDENT y1
  /METHOD=ENTER x1  .
*Create a version of the variable with all points at values 2 and 4.
COMPUTE X2=2.
IF X1 GT 3 X2=4.
COMPUTE Y2= X2 + E.
EXECUTE.
GRAPH
  /HISTOGRAM(NORMAL)=x2 .
REGRESSION
  /DESCRIPTIVES MEAN STDDEV CORR SIG N
  /MISSING LISTWISE
  /STATISTICS COEFF OUTS R ANOVA
  /CRITERIA=PIN(.05) POUT(.10)
  /NOORIGIN
  /DEPENDENT y2
  /METHOD=ENTER x2  .
*CREATE A MORE EXTREME EXAMPLE.
COMPUTE X3=1.
IF X1 GT 3 X3=5.
COMPUTE Y3= X3 + E.
EXECUTE.
GRAPH
  /HISTOGRAM(NORMAL)=x3 .
REGRESSION
  /DESCRIPTIVES MEAN STDDEV CORR SIG N
  /MISSING LISTWISE
  /STATISTICS COEFF OUTS R ANOVA
  /CRITERIA=PIN(.05) POUT(.10)
  /NOORIGIN
  /DEPENDENT y3
  /METHOD=ENTER x3  .