Creating multiple regression terms

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Creating multiple regression terms

Anton-24
Dear SPSS list members,

I have been using a statistical consultant to help me with a couple of
multiple regression runs and I'm just a bit skeptical about the way the data
were set up.

I have essentially two independent variables: Group (coded 0 or 1) and
number of days before or after a cut-off date (range = -365 to +365).  These
two variables are used to create a Group-by-Days interaction term, quadratic
and cubic Days terms, and quadratic and cubic Group-by-Days interactions.  I
am using test scores as dependent variables.

Is it proper to code Group 0 or 1 when Group is going to be used to create
interaction terms.  Is it okay to use negative numbers when creating squared
and cubed effects and their subsequent interactions?

Any help will assist me in sleeping more soundly at night!


The actual values for each of the variables for 4 cases are shown below.


1  = Case ID
0  = Group
2  = Number of days to cut-off date
0  = Group BY Number of days to cut-off date Interaction
4  = Number of days to cut-off date (Squared)
0  = Group BY Number of days to cut-off date Interaction (Squared)
8  = Number of days to cut-off date (Cubed)
0  = Group BY Number of days to cut-off date Interaction (Cubed)


2 = Case ID
0 = Group
-10 = Number of days to cut-off date
0 = Group BY Number of days to cut-off date Interaction
100 = Number of days to cut-off date (Squared)
0 = Group BY Number of days to cut-off date Interaction (Squared)
-1000 = Number of days to cut-off date (Cubed)
0 = Group BY Number of days to cut-off date Interaction (Cubed)


3 = Case ID
1 = Group
15 = Number of days to cut-off date
15 = Group BY Number of days to cut-off date Interaction
225 = Number of days to cut-off date (Squared)
225 = Group BY Number of days to cut-off date Interaction (Squared)
3375 = Number of days to cut-off date (Cubed)
3375 = Group BY Number of days to cut-off date Interaction (Cubed)


4 = Case ID
1 = Group
-7 = Number of days to cut-off date
-7 = Group BY Number of days to cut-off date Interaction
49 = Number of days to cut-off date (Squared)
49 = Group BY Number of days to cut-off date Interaction (Squared)
-343 = Number of days to cut-off date (Cubed)
-343 = Group BY Number of days to cut-off date Interaction (Cubed)
Reply | Threaded
Open this post in threaded view
|

Re: Creating multiple regression terms

Swank, Paul R
Go to sleep. It's fine. The only question is what is the nature of the
relation of number of days with the outcome. Have you graphed it
(scatterplot) to see if the curvature parameters are necessary?


Paul R. Swank, Ph.D. Professor
Director of Reseach
Children's Learning Institute
University of Texas Health Science Center-Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Anton
Sent: Monday, March 26, 2007 5:12 PM
To: [hidden email]
Subject: Creating multiple regression terms

Dear SPSS list members,

I have been using a statistical consultant to help me with a couple of
multiple regression runs and I'm just a bit skeptical about the way the
data were set up.

I have essentially two independent variables: Group (coded 0 or 1) and
number of days before or after a cut-off date (range = -365 to +365).
These two variables are used to create a Group-by-Days interaction term,
quadratic and cubic Days terms, and quadratic and cubic Group-by-Days
interactions.  I am using test scores as dependent variables.

Is it proper to code Group 0 or 1 when Group is going to be used to
create interaction terms.  Is it okay to use negative numbers when
creating squared and cubed effects and their subsequent interactions?

Any help will assist me in sleeping more soundly at night!


The actual values for each of the variables for 4 cases are shown below.


1  = Case ID
0  = Group
2  = Number of days to cut-off date
0  = Group BY Number of days to cut-off date Interaction
4  = Number of days to cut-off date (Squared) 0  = Group BY Number of
days to cut-off date Interaction (Squared)
8  = Number of days to cut-off date (Cubed) 0  = Group BY Number of days
to cut-off date Interaction (Cubed)


2 = Case ID
0 = Group
-10 = Number of days to cut-off date
0 = Group BY Number of days to cut-off date Interaction 100 = Number of
days to cut-off date (Squared) 0 = Group BY Number of days to cut-off
date Interaction (Squared) -1000 = Number of days to cut-off date
(Cubed) 0 = Group BY Number of days to cut-off date Interaction (Cubed)


3 = Case ID
1 = Group
15 = Number of days to cut-off date
15 = Group BY Number of days to cut-off date Interaction
225 = Number of days to cut-off date (Squared)
225 = Group BY Number of days to cut-off date Interaction (Squared)
3375 = Number of days to cut-off date (Cubed)
3375 = Group BY Number of days to cut-off date Interaction (Cubed)


4 = Case ID
1 = Group
-7 = Number of days to cut-off date
-7 = Group BY Number of days to cut-off date Interaction
49 = Number of days to cut-off date (Squared)
49 = Group BY Number of days to cut-off date Interaction (Squared)
-343 = Number of days to cut-off date (Cubed)
-343 = Group BY Number of days to cut-off date Interaction (Cubed)