What kind of analysis is this?

[Brock-15]

Hi everyone,

In my head, I know what I want to do, but I am not quite sure how to go
about it.  First, let me explain my data.  I have a list of competitors
that my prospect lists in their profile.  Eventually they may select us,
they may select one of the listed competitors, or select someone else
entirely.  Right now the data are set up one record per prospect, with up
to 6 columns for competitors (we may be listed in this list of 6).  What I
am looking to do is probably similar to examples of social network
analysis (I think).  I want to see, visually, where all of the competitors
listed fall and how "similar" they are.

I guess another way to look at it would be similar to the challenge
offered by Netflix (I believe they called it collaborative filtering).  If
I have my list, what other competitors could we assume this prospect has
looked at.  The goal ultimately would be to use the information up-front
to get a better understanding of the likelihood of selecting us,
sensitivity to promotions, etc.

Trust me, I know this is an extremely open question, but what techniques
are available, especially in SPSS?  I have even seen examples of the Text
Analysis Software and thought about doing word associations based on the
competitors listed, but I am not sure that would work.  Lastly, how would
my data need to be set up, and what additional information should I
include in my dataset.

Many thanks for any light you can shine on this subject.

~ Brock

[Philip_Moore]

Brock,

I'm sure there is a really cool and sophisticated way to do this that
doesn't require what I'm going to propose (which I too am eager to hear
about).  I presume the competitors can appear in any of the six columns.
If the column in which they appear is meaningful (like which competitor
they chose first from a list goes in the first column), then my proposed
solution is too simple and will lose that information.

Step 1 - define the six columns as a multiple response variable, then do a
frequency to get a comprehensive list of all the competitors included in
the data
Step 2 - inialize a new variable for each competitor to 0  (there is
probably some cool vector and looping syntax to do this but I don't know
it)
Step 3 - execute an if statement for the six columns to change the value of
the variable to 1
With you data in this format you can do correlation tables & logistic
regression (and probably a bunch of other techniques I haven't learned yet)
to find the relationships you're looking for.

Good luck,

Philip Moore
Market Research Manager
CarMax, the Auto Superstore
named America's Most Admired Auto Retailer
by Fortune Magazine, 2006



             [hidden email]
             du
             Sent by:                                                   To
             "SPSSX(r)                 [hidden email]
             Discussion"                                                cc
             <SPSSX-L@LISTSERV
             .UGA.EDU>                                             Subject
                                       What kind of analysis is this?

             01/30/2007 09:20
             AM


             Please respond to
             [hidden email]
                    du






Hi everyone,

In my head, I know what I want to do, but I am not quite sure how to go
about it.  First, let me explain my data.  I have a list of competitors
that my prospect lists in their profile.  Eventually they may select us,
they may select one of the listed competitors, or select someone else
entirely.  Right now the data are set up one record per prospect, with up
to 6 columns for competitors (we may be listed in this list of 6).  What I
am looking to do is probably similar to examples of social network
analysis (I think).  I want to see, visually, where all of the competitors
listed fall and how "similar" they are.

I guess another way to look at it would be similar to the challenge
offered by Netflix (I believe they called it collaborative filtering).  If
I have my list, what other competitors could we assume this prospect has
looked at.  The goal ultimately would be to use the information up-front
to get a better understanding of the likelihood of selecting us,
sensitivity to promotions, etc.

Trust me, I know this is an extremely open question, but what techniques
are available, especially in SPSS?  I have even seen examples of the Text
Analysis Software and thought about doing word associations based on the
competitors listed, but I am not sure that would work.  Lastly, how would
my data need to be set up, and what additional information should I
include in my dataset.

Many thanks for any light you can shine on this subject.

~ Brock

[Brock-15]

Thanks for getting back to me.  For some analyses, I would not want to
lose the ordering of the competitors.  I have found that the likelihood of
selecting us decreases as we are listed as a 3rd, 4th, 5th,etc..  However,
my goal is to simply find some association between the competitors listed.
 I figure that if Competitor A is listed, there has to be some technique
that says Competitors B, C, and D are often also selected.  In addition, I
figure that if Competitor Z is listed but never with Competitor A, that
these two organizations are highly different.

My data file has approx 1600 cases and a few hundred "competitors."
Clearly I am trying to simply my example by putting this in a business
context.  I would imagine this would be easier if I had ratings, but all I
am working with is competitors listed, and some various data about the
respondent.

Thanks again,

~ Brock