Date: Fri, 14 Feb 2003 11:26:00 -0500
Reply-To: "Catherine E. Brawner" <firstname.lastname@example.org>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Catherine E. Brawner" <email@example.com>
Subject: Re: Graphing Multiple "Weird" Variables on One Graph
Content-Type: text/plain; charset="us-ascii"; format=flowed
I can't give you syntax, but I think I can help you conceptually with your
problem since I have faced the same thing. What you need to do is create a
multiple response set named pets_n_fruits that contains monkeys, cats,
apples, and bananas. In 11.5 in the GUI, go to analyze/multiple
response/define sets. Put your variables in there and select dichotomy and
1 as the selection variable. The you can run multiple response frequencies
(same menu). Be sure to save the syntax because the multiple response sets
tend to disappear when you close SPSS (not sure why). What I do from here
is export the output into excel and make graphs from there.
You can also do this from analyze/tables/multiple response sets and make
tables instead which tend to be prettier. Again, define the sets and make
the table you want, then export into excel to make graphs. The multiple
response sets in tables don't seem to disappear like they do in multiple
response and the tables are a bit easier to work with.
While this is a multi-step process to solve your problem, the export to
excel works very well in 11.5 and making the graphs is a cinch from there.
If you don't have 11.5 and really need to do this, it is worth the money to
At 01:27 PM 2/14/03 +0100, you wrote:
>I have quite an interesting problem,and if you bear with the explanation,
>perhaps you can help me!
>I have data imported from an ACCESS database. If you are familiar with
>ACCESS, you may know that there are certain variables that can be either
>"Yes/No". When you import these variables into SPSS, they come out as 1s
>and 0s. I have replaced the 0s with "System Missing"s.
>I now have 18 variables that are simply columns of empty cells and 1s. I
>want to analyse these 18 variables. I would like to create a graph which
>graphs each of these 18 variables along the x axis. The complicated thing
>is that I want to graph 18 different variables in the same graph, and that
>I want the total for each of these variables as a percent of all cases in
>the data set.
>For example, each of these 18 variables represent whether a journal
>article discussed a certain item within the article, with a Yes or No (I
>am simplifying the situation for purposes of explanation). If a journal
>article happened to discuss apples, bananas, monkeys and cats, then each
>of these variables (assuming that there were apple, banana, monkey and cat
>variables!) would receive a value of 1 for this case. The rest of the 18
>variables for this case would be "system missing". I wish to graph the
>percent of all papers which addressed each of these items (for example,
>if 2 of the 4 articles discussed apples, the bar for "Apples" would be at
>the 50% mark). In other words, I wish to graph each of the "items"
>(monkeys, cats, bananas, and apples) along the x axis, and the percent of
>all papers in the data set that discussed this item.
>This may sound simple but I have found no way to accomplish this! I
>beleive this is difficult because it involves summing up the values across
>all cases and then mapping out 18 variables onto one graph. SPSS has
>rejected my every effort to accomplish this. I would appreciate the
>knowledge of the group on this matter.
>Thank you for reading this long explanation and for any advice you may have!
Catherine E. Brawner, Ph.D.
Research Triangle Educational Consultants
6316 Lakeland Drive
Raleigh, NC 27612