Error # 17862: why will SPSS sometimes not graph variables from different conditions?

[Emma]

I am trying to have SPSS graph the mean of certain variables. These
variables only have data one out of every 4 lines, as there were 4
conditions that my subjects were in. When I select variables from
different conditions, SPSS will not graph the variables, giving me the
below error message instead:

>Error # 17862
>There are not enough cases with valid values available for GRAPH to
>process.
>This command not executed.
>Chart not produced: Bar of mean(fpt1_a fct1_c spt1_a sct1_c)

It seems that it will not produce a graph that makes it calculate means
from different conditions. i.e. if variables A and B were completed by
different sets of subjects, it won't graph the means of variables A and
B; if they were completed by the same subjects, it will.

Is there any way to get around this?

In case it is a memory issue, can you tell me how to increase memory
for SPSS in Windows XP? (I have tried looking at the "properties," but
can't find a "memory" tab that my XP "help" says should be there...)

Thank you very much for any help!

yours, Emma Buchtel

[Bruce Weaver]

I'm not clear on how exactly your data are structured. Each subject is
appearing in 4 conditions, right? A lot of folks would have one row
per subject with the 4 condition scores in 4 separate columns. Your
use of 4 different variable names above suggests that you may have it
this way. But then you talk about different sets of subjects and every
4 lines, and I start losing the plot. Here are a couple of examples
that may help you make some progress.

* ----- Start of syntax ------ .

* Each subject in 4 conditions, one row per subject.

data list free / id a b c d (5f5.0).
begin data.
1 4 5 4 8
2 5 4 3 7
3 2 6 8 2
4 7 8 3 5
5 6 2 7 4
end data.

GRAPH
/BAR(SIMPLE)= MEAN(a) MEAN(b) MEAN(c) MEAN(d)
/MISSING=LISTWISE
/TITLE= 'Your title here'.

* Now restructure to have 4 rows per ID,
* and all the scores in one column.

VARSTOCASES
/MAKE score FROM a to d
/INDEX = cond(4)
/KEEP = id
/NULL = drop.

* Add variable labels for COND if you want them.

GRAPH
/BAR(SIMPLE)=MEAN(score) BY cond
/TITLE= 'Title here'.

* ----- End of syntax ------ .

HTH.

--
Bruce Weaver
bwe...@lakeheadu.ca
www.angelfire.com/wv/bwhomedir

[Emma]

Thanks very much for your advice!

I didn't explain myself very clearly, however: each subject only did
one of four conditions. So the data looks kinda like this:
# c1-1 c1-2 c2-1 c2-2
1 34 78
2 39 29
3 24 39
4 29 28
5 10 90

and so on. (in this case, for simplicity's sake, I'm making it look
like there were 2 conditions, but if there's a solution for the above
data, it'll work for my 4-condition data...)

When I ask SPSS to graph, say, the means of c1-1 and c1-2 together,
that's no problem. But if I ask SPSS to graph the means of C2-1 and
c2-2 together, it gives me that error message.

Looking at your syntax again, I'm realizing that I've never tried
putting all the data in one column and doing the "graph score by
condition" thing. Would that solve my problem? I will try to rearrange
the data, but if you know an easy way to do it could you tell me?

Thanks again! I really appreciate the help! -Emma

[Emma]

Sorry, I again made an error in my explanation. Graphing c2-1 and c2-2
is also not a problem. The problem occurs if I want to graph, e.g.,
c1-1 and c2-1 together.

If I can do this without rearranging the data that would be great, as I
already have a lot of analyses saved in syntax with the old
arrangement..!

Thanks again, Emma

p.s. also, please ignore the incorrect spacing of the data and variable
names in my example. I thought the posting would be in courier...

[Bruce Weaver]

Okay, if each subject appears in one and only one of the 4 conditions,
you need to structure your data like this:

a b score

1 1 34
1 2 78
2 1 39
2 2 29
1 1 24
1 2 39
2 1 29
2 2 28
1 1 10
1 2 90

The first column (a) shows 1 and 2 for your c1 and c2 respectively; and
the second column (b) shows 1 and 2 for your -1 and -2. Then you can
show the means for all 4 cells in the design in one graph with a
clustered bar chart. You can define the clusters by either a or b,
like this:

GRAPH
/BAR(GROUPED)=MEAN(score) BY b BY a
/TITLE= 'Title here'.

GRAPH
/BAR(GROUPED)=MEAN(score) BY a BY b
/TITLE= 'Title here'.

To restructure your data from its current state, use VARSTOCASES, like
this:

VARSTOCASES
/MAKE score FROM c11 to c22
/INDEX = a(2) b(2)
/drop = id
/NULL = drop.

This syntax assumes that the variable you used to number the rows was
called ID. You can change it if necessary (on the /drop line above).

[Emma]

Thank you so much! I really appreciate your taking the time to help. I
can now graph everything properly!

Thanks, Emma