How to tabulate combinations of multiple responses

[Frank Furter]

I have a list of about 20 magazines that people may be reading. Since noone in the dataset reads more than 6 magazines regularly, I have stored them in 6 variables (say, mag1 through mag6) as multiple responses (of course, I could also store them as multiple dichotomies, one variable for each magazine, if this would make things easier).

In one analysis I am not interested in the percentage of subjects who read a particular magazine, but in the frequency of combinations of magazines that people read regularly. Naturally, I would like to avoid having to look at each possible combination of 1-6 out of 20 magazines, as this would be a very large number and most of them have never occurred in the data.

Is it possible to identify the combinations that have actually occurred automatically - and how do I do this?

[David Marso]

Andreas,
SPSS has had a MULTIPLE RESPONSE procedure for decades.
Poke around a bit and you will find it.
The topic has also shown up in a few threads this past week.
If you want to enumerate the various pairs, triplets, quads etc.  
The best way would be to write code in MATRIX.
David

Andreas Voelp wrote

I have a list of about 20 magazines that people may be reading. Since noone in the dataset reads more than 6 magazines regularly, I have stored them in 6 variables (say, mag1 through mag6) as multiple responses (of course, I could also store them as multiple dichotomies, one variable for each magazine, if this would make things easier).

In one analysis I am not interested in the percentage of subjects who read a particular magazine, but in the frequency of combinations of magazines that people read regularly. Naturally, I would like to avoid having to look at each possible combination of 1-6 out of 20 magazines, as this would be a very large number and most of them have never occurred in the data.

Is it possible to identify the combinations that have actually occurred automatically - and how do I do this?

[Vik Rubenfeld]

You may also want to look at doing a Cluster Analysis to see which sets of magazines cluster together. The multiple dichotomies approach to variable structure may be easier to work with for this purpose.

On Jan 20, 2013, at 10:49 AM, Andreas Voelp wrote:

> I have a list of about 20 magazines that people may be reading. Since noone
> in the dataset reads more than 6 magazines regularly, I have stored them in
> 6 variables (say, mag1 through mag6) as multiple responses (of course, I
> could also store them as multiple dichotomies, one variable for each
> magazine, if this would make things easier).
>
> In one analysis I am not interested in the percentage of subjects who read a
> particular magazine, but in the frequency of combinations of magazines that
> people read regularly. Naturally, I would like to avoid having to look at
> each possible combination of 1-6 out of 20 magazines, as this would be a
> very large number and most of them have never occurred in the data.
>
> Is it possible to identify the combinations that have actually occurred
> automatically - and how do I do this?
>
>
>
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/How-to-tabulate-combinations-of-multiple-responses-tp5717540.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[David Marso]

Indeed,
Cluster will reveal the 'pairwise' associations but not enumerate them nor reveal higher order associations.
The MD format would be required in either case as the MR format would simply result in GIGO.
To map MR to MD consider.
*Map*.
VECTOR MD(20) / MR=MR1 TO MR6.
RECODE MD (ELSE=0).
LOOP #=1 TO 6.
COMPUTE MD(MR(#))=1.
END LOOP.

*Concat*.
STRING MDStr (A20).
LOOP #=1 TO 20.
COMPUTE SUBSTR(MDStr(#,1))=STRING(MD(#),F1).
END LOOP.

*Enumerate*.
FREQ MDStr.

Vik Rubenfeld wrote

You may also want to look at doing a Cluster Analysis to see which sets of magazines cluster together. The multiple dichotomies approach to variable structure may be easier to work with for this purpose.

On Jan 20, 2013, at 10:49 AM, Andreas Voelp wrote:

> I have a list of about 20 magazines that people may be reading. Since noone
> in the dataset reads more than 6 magazines regularly, I have stored them in
> 6 variables (say, mag1 through mag6) as multiple responses (of course, I
> could also store them as multiple dichotomies, one variable for each
> magazine, if this would make things easier).
>
> In one analysis I am not interested in the percentage of subjects who read a
> particular magazine, but in the frequency of combinations of magazines that
> people read regularly. Naturally, I would like to avoid having to look at
> each possible combination of 1-6 out of 20 magazines, as this would be a
> very large number and most of them have never occurred in the data.
>
> Is it possible to identify the combinations that have actually occurred
> automatically - and how do I do this?
>
>
>
> --
> View this message in context: http://spssx-discussion.1045642.n5.nabble.com/How-to-tabulate-combinations-of-multiple-responses-tp5717540.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Frank Furter]

... thanks, very helpfulk

Andreas