Date:         Tue, 23 May 2006 11:03:36 -0400
Reply-To:     "Feinstein, Zachary" <ZFeinstein@HarrisInteractive.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Feinstein, Zachary" <ZFeinstein@HarrisInteractive.com>
Subject:      Re: Trying to Create Histogram in SPSS That Overlays Breaks in
              Variable - Is GPL The Way?
Comments: To: "Beadle, ViAnn" <viann@spss.com>
Content-Type: text/plain; charset="us-ascii"

Thank you ViAnn.  The whole reason I brought up the issue of bar-charts
was so that I could stack the vertical lines of the bins next to each
other.  For example, using the example you provided below, there would a
bin for low ages and multiple bars for the (let's say) three regions
there, then the next bin would have three vertical bars for the three
regions, and so on.  Currently, the code below makes separate histograms
for each value of region.  I want/need them combined to all be in one,
if that' s possible.
 
Two more questions to go with this.
 
1.    How do we control the bin size?
2.    Can I display a percentage instead of a COUNT.
 
Thanks again.
 
Zachary

________________________________

From: Beadle, ViAnn [mailto:viann@spss.com] 
Sent: Monday, May 22, 2006 9:30 PM
To: Feinstein, Zachary; SPSSX-L@LISTSERV.UGA.EDU
Subject: RE: Trying to Create Histogram in SPSS That Overlays Breaks in
Variable - Is GPL The Way?


Do you want a bar chart of counts or a histogram in which interval level
variables are "binned". The primary difference between the two is 1) the
x axis has an interval scale in a histogram while a bar chart of counts
usually doesn't; 2) a histogram tries to compute nice intervals for
binning while a bar chart of counts provides a bar for every observed
value of the x axis variable.
 
You can get side by side histograms through GPL or through a basic GRAPH
command. Let's start with your basic graph command: 
 
GRAPH /HISTOGRAM=age/ PANEL COLVAR=region COLOP=CROSS .
 
The GPL is a bit more complicated:
GGRAPH

/GRAPHDATASET NAME="graphdataset" VARIABLES=age region MISSING=LISTWISE

REPORTMISSING=NO

/GRAPHSPEC SOURCE=INLINE.

BEGIN GPL

SOURCE: s=userSource(id("graphdataset"))

DATA: age=col(source(s), name("age"))

DATA: region=col(source(s), name("region"), unit.category())

GUIDE: axis(dim(1), label("Age of Respondent"))

GUIDE: axis(dim(2), label("Frequency"))

GUIDE: axis(dim(3), label("Region of the U.S"))

ELEMENT: interval(position(summary.count(bin.rect(age*1*region))),
shape.interior(

shape.square))

END GPL.

 

The break variable is crossed with the histgram variable. The "1" is
required to stand in for the y-axis in the algebra

because summary.count only implicitly creates the second dimension.
WARNING--this was tested with a new version

of GPL and might require some tweaking on your part.

 

________________________________

From: SPSSX(r) Discussion on behalf of Feinstein, Zachary
Sent: Mon 5/22/2006 4:19 PM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: Trying to Create Histogram in SPSS That Overlays Breaks in
Variable - Is GPL The Way?



I have had very good luck utilizing the listserv for figuring how to do
some nifty scatterplots.  Thanks to all who have replied.  Now I need to
build a histogram for a single variable but I would like multiple
histograms for each of my breaks overlaid on top of each other.

The code below builds a histogram but it does not group the categories
of the x-axis.  I would need to figure how to do that.  Then, assuming I
have a three-category variable called break1, how would I do three
histograms in one so that they show up side-by-side?  Is this even
doable or would I have to create a bar-chart of sorts?  Perhaps break
the variables in the datafile then do a simple bar chart?  Would greatly
prefer a histogram solution if one exists.

Thanks.

Zachary
zfeinstein@harrisinteractive.com
GGRAPH

/GRAPHDATASET NAME = "hist1" VARIABLES = loy_var5 COUNT()

/GRAPHSPEC SOURCE = INLINE.

BEGIN GPL

SOURCE: s = userSource(id("hist1"))

DATA: loy_var5 = col(source(s), name("loy_var5"), unit.category())

DATA: count = col(source(s), name("COUNT"))

GUIDE: axis(dim(1), label("Loyalty Level"))

GUIDE: axis(dim(2), label("Count"))

ELEMENT: interval(position(loy_var5 * count))

END GPL.