11 January 2012

First thought question. Run the below listed code with and without
stepwise option.

b34sexec options ginclude('greene08.mac') member(table25_1); b34srun;
b34sexec regression residuala stepwise;
model majordrg=age income exp_inc avgexp ownrent selfempl
               depndt inc_per cur_add major active;
b34srun;


Questions.

1.  Is avgexp significant for complete model?

2. When does avgexp enter the stepwise model?  
   (Help: avgexp is variabl # 6) 

3. If avgexp once was significant, what variable "killed" its 
significance?  What does this tell you?  How might this logic 
exercise help you in research in a bias case?

**************************************************************************
19 January 2012

1. First assingment due 31 January since no one objected.

2. For 10 extra points use matlab or R or SAS

3. L1  and MINIMAX are on the call olsq command



call olsq(y x1 x2 :l1 :minimax :print);


See chapter 2

*********************************************************************
31 January 2012

b34sexec matrix;
v1=vector(:1 2 3 4);
pp=v1*v1;
call print(v1,pp);
m1=matrix(4,1:v1);
call print(m1,m1*transpose(m1),transpose(m1)*m1);
b34srun;


The code listed above illustrates that b34s knows how to multiply two 
vectors or a vector and a matrix and does not require you to transpose a 
vector.

**************************************************************************