procedure for experimental design

procedure for experimental design

Post by Scott Dixo » Tue, 08 Sep 1998 04:00:00



A colleague asked for my help with his thesis project.  He performed a
simple experiment.  He pretested two classes using the same test.  Then he
gave one class a special set of lectures, and gave the other class the
lectures he always uses.  Finally, he gave both classes the same posttest.
He wants to see if the special lectures had an effect on the posttest of
the experimental group compared with the posttest of the control group.  I
am having trouble coming up with the proper SPSS procecure.

The design is:

Group                Observation-1    Stimulus      Observation-2

experimental class    pretest        new-lectures     posttest

control class         pretest        old-lectures     posttest

The two groups were not randomly assigned, nor matched along any variable.
However, the groups were not allowed to pre-select themselves into either
class.  The class contents were similar in every way except the set of new
lectures.  The experimental group did not know that they were being exposed
to new lectures.

I am embarrassed to say that I cannot figure out what SPSS procedure to
perform.  The closest I can get is "Independent-Samples T Test".  Here, I
used the continuous variables "Pretest" and "Posttest" by "Classes [groups
1,2]".  But it doesn't seem to say if the improvement between pretest and
posttest is significantly greater between the two groups.  Is that
something I have to do?  Or is there a procedure that calculates that for me?

I would be grateful if someone could point me in the right direction.  I
will gladly pour over the books if someone would give me a push in the
general direction.

Thank you very much.

Scott Dixon

 
 
 

procedure for experimental design

Post by William B. War » Tue, 08 Sep 1998 04:00:00



> The design is:

> Group                Observation-1    Stimulus      Observation-2

> experimental class    pretest        new-lectures     posttest

> control class         pretest        old-lectures     posttest

> The two groups were not randomly assigned, nor matched along any variable.
> However, the groups were not allowed to pre-select themselves into either
> class.  The class contents were similar in every way except the set of new
> lectures.  The experimental group did not know that they were being exposed
> to new lectures.

The situation that you describe appears best modeled using
"analysis of covariance."  I say "best" because the situation does not
exactly meet all of the assumptions... You don't have random
assignment, and the treatment was administer to groups, which means
that the errors are probably not independent.  But what the heck...

Make sure that you have a grouping variable that is coded "1" and "2".
Within GLM using either Simple Factorial or General Factorial, declare
"posttest" to be the DV, "group" to be the factor, and declare "pretest"
as the covariate.  You should also check to see if the two group
regression lines (post on pre) are parallel...  You can check that
visually by looking at a scatterplot in which you request that "group" be
used as a marker... Or you can test it statistically within General
Factorial by customizing your design and creating the interaction term
between pretest and group.

__________________________________________________________________________
William B. Ware, Professor and Chair                 Psychological Studies
CB# 3500                                                      in Education
University of North Carolina                         PHONE  (919)-962-7848
Chapel Hill, NC      27599-3500                      FAX:   (919)-962-1533

__________________________________________________________________________

 
 
 

procedure for experimental design

Post by Michael A. Schoenfiel » Tue, 08 Sep 1998 04:00:00


This may seem a stupid basic question, but the validity of the test is at
stake here. Did your friend assign students randomly to the two classes? If
they were not randomly assigned, I can foresee problems which may deem the
experimental design. at best questionable.

Michael

Michael A. Schoenfield & Associates, Ltd.
2637 Mason Street
Madison, WI 53705-3709

(608) 238-6121     Voice
(608) 233-2507     Fax

http://www.angelfire.com/biz/maschoen/     WWW Site

-----Original Message-----

Newsgroups: bit.listserv.spssx-l

Date: Monday, September 07, 1998 8:35 AM
Subject: procedure for experimental design

>A colleague asked for my help with his thesis project.  He performed a
>simple experiment.  He pretested two classes using the same test.  Then he
>gave one class a special set of lectures, and gave the other class the
>lectures he always uses.  Finally, he gave both classes the same posttest.
>He wants to see if the special lectures had an effect on the posttest of
>the experimental group compared with the posttest of the control group.  I
>am having trouble coming up with the proper SPSS procecure.

>The design is:

>Group                Observation-1    Stimulus      Observation-2

>experimental class    pretest        new-lectures     posttest

>control class         pretest        old-lectures     posttest

>The two groups were not randomly assigned, nor matched along any variable.
>However, the groups were not allowed to pre-select themselves into either
>class.  The class contents were similar in every way except the set of new
>lectures.  The experimental group did not know that they were being exposed
>to new lectures.

>I am embarrassed to say that I cannot figure out what SPSS procedure to
>perform.  The closest I can get is "Independent-Samples T Test".  Here, I
>used the continuous variables "Pretest" and "Posttest" by "Classes [groups
>1,2]".  But it doesn't seem to say if the improvement between pretest and
>posttest is significantly greater between the two groups.  Is that
>something I have to do?  Or is there a procedure that calculates that for
me?

>I would be grateful if someone could point me in the right direction.  I
>will gladly pour over the books if someone would give me a push in the
>general direction.

>Thank you very much.

>Scott Dixon


 
 
 

1. Experimental Design (PROC PLAN)

Hi.  Does anyone know of any good reference articles or books on
experimental design (randomized complete block, incomplete block, etc.)
using PROC PLAN.  I know that SAS has a QC procedure for doing experimental
designs, but I am only interested in using PROC PLAN.  I have also read the
on-line documentation for PROC PLAN, but I am interested in more detailed
descriptions and examples.

2. Can the netBook/7 handle a 800*600 resolution?

3. Position Available-Ph.D Statistics w/Experimental Design Experience

4. CD Drive audio output

5. Experimental Design

6. Mechanical/Electronic Design and Drafting (Pro/E, Solidworks, AutoCAD)

7. Help - Experimental design problems

8. DEC LN03 Script Printer acting strange

9. Experimental Design Text Advice

10. procedure within a procedure

11. Mapping performed procedure codes to RIS "billing" scheduled procedure codes

12. type(sin,procedure) = true, type (sin@sin, procedure)=?

13. type(sin,procedure) = true, type(sin@sin, procedure)=?