Fw: R-matrix

[jacqueline london]

-----
Subject: R-matrix

Hi all.
I would like to do factor analysis of a questionnaire that I have. I want to plot a scatterplot graph first to see how things are, before I do the matrix. There are 30 items in the questionnaire and I have 155 subjects. However I tried to plot a simple scatterplot using only the ist and 2nd items. I got a graph with only about 22 point with no clusters or even 2 point close. I do not know what to do. Shouldn't I see the correlation between item 1 and item 2 for all 155 participants? Would someone please help?

[Art Kendall]

What is the range of legitimate values for the variables? Are there likely to be many cases that have a particular pair of values?
For example, if you have items that have a 1 to 5 response scale,  the are only 25 possible locations in a scatterplot that cases can be located at. Also with 25 possible locations it is possible that some of the 25 possible locations have no cases.

Please describe the variables more fully.

Also some of the scatterplot graphics allow you to "Jitter". 
<help> <search> "jitter"

Art Kendall
Social Research Consultants



On 7/21/2010 10:06 PM, jacqueline london wrote:

-----
Subject: R-matrix

Hi all.
I would like to do factor analysis of a questionnaire that I have. I want to plot a scatterplot graph first to see how things are, before I do the matrix. There are 30 items in the questionnaire and I have 155 subjects. However I tried to plot a simple scatterplot using only the ist and 2nd items. I got a graph with only about 22 point with no clusters or even 2 point close. I do not know what to do. Shouldn't I see the correlation between item 1 and item 2 for all 155 participants? Would someone please help?

===================== 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

[Art Kendall]

On the simplest scatterplots, cases would be exactly at the same point.  See "jitter" in <help>. Jittering moves points out from behind one another.

The syntax below simulates throwing 2 dice. (values are 1 to 6 rather than 1 to 5.)
A crosstab is analogous to a scatter plot except that the points are counts rather than simple elements.
Then there are 3 scatterplots of the same data with no jitter, and with 2 kinds of collision modifiers.

* throw 2 dice.
SET SEED=20100723.
NEW FILE.
INPUT PROGRAM  .
LOOP id=1 to 155.
COMPUTE die1=RND(RV.UNIFORM(.5,6.5)).
COMPUTE die2=RND(RV.UNIFORM(.5,6.5)).
END CASE.
END LOOP.
END FILE.
END INPUT PROGRAM.
FORMATS DIE1 DIE2 (F1).
VARIABLE LEVEL die1 die2 (nominal).
CROSSTAB TABLES= DIE1 BY DIE2 /CELLS=count.
CROSSTAB TABLES= DIE1 BY DIE2 /CELLS=ALL.
* scatterplot without jitter.
GGRAPH
  /GRAPHDATASET NAME="graphdataset"
    VARIABLES=die2[LEVEL=ratio] die1[LEVEL=ratio]
    MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=VIZTEMPLATE(NAME="Scatterplot"[LOCATION=LOCAL]
    MAPPING( "y"="die2"[DATASET="graphdataset"] "x"="die1"[DATASET="graphdataset"]))
    VIZSTYLESHEET="Traditional"[LOCATION=LOCAL]
    LABEL="Scatterplot: die1-die2"
    DEFAULTTEMPLATE=NO.
*scatterplot with dodge.symmetric jitter.
* Chart Builder.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=die1 die2 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: die1=col(source(s), name("die1"), unit.category())
  DATA: die2=col(source(s), name("die2"), unit.category())
  GUIDE: axis(dim(1), label("die1"))
  GUIDE: axis(dim(2), label("die2"))
  ELEMENT: point.dodge.symmetric(position(die1*die2))
END GPL.
*scatterplot with jitter.normal.
* Chart Builder.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=die1 die2 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: die1=col(source(s), name("die1"), unit.category())
  DATA: die2=col(source(s), name("die2"), unit.category())
  GUIDE: axis(dim(1), label("die1"))
  GUIDE: axis(dim(2), label("die2"))
  ELEMENT: point.jitter.normal(position(die1*die2))
END GPL.

Art Kendall
Social Research Consultants


On 7/22/2010 11:49 PM, jacqueline london wrote:

The range of values is 1 to 5 : strongly agree, agree, neutral, disagree and strongly disagree. Yes I believe from 155 there would be a significant number of cases that would have a particular pair of values. Yes there were some points with no cases, but what puzzled me is that all had single plots, how come not even 1 had 2 plots. But I think I understand what you are saying with only 25 possible locations and the 2 items that I chose it is quite possible. I guess the simple scatterplot is just to give you an idea.  I will try other pairs and see what happens. Thanks.

---

From: Art Kendall [hidden email]
To: jacqueline london [hidden email]
Cc: [hidden email]
Sent: Thu, July 22, 2010 4:14:03 PM
Subject: Re: [SPSSX-L] Fw: R-matrix

What is the range of legitimate values for the variables? Are there likely to be many cases that have a particular pair of values?
For example, if you have items that have a 1 to 5 response scale,  the are only 25 possible locations in a scatterplot that cases can be located at. Also with 25 possible locations it is possible that some of the 25 possible locations have no cases.

Please describe the variables more fully.

Also some of the scatterplot graphics allow you to "Jitter". 
<help> <search> "jitter"

Art Kendall
Social Research Consultants



On 7/21/2010 10:06 PM, jacqueline london wrote:

-----
Subject: R-matrix

Hi all.
I would like to do factor analysis of a questionnaire that I have. I want to plot a scatterplot graph first to see how things are, before I do the matrix. There are 30 items in the questionnaire and I have 155 subjects. However I tried to plot a simple scatterplot using only the ist and 2nd items. I got a graph with only about 22 point with no clusters or even 2 point close. I do not know what to do. Shouldn't I see the correlation between item 1 and item 2 for all 155 participants? Would someone please help?

===================== 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