Re: SPSSX-L Digest - 28 Jul 2007 to 29 Jul 2007 (#2007-211)

[Chelminski, Iwona]

Hi Group,

I've got a suggestions from some that SPSS v. 15 is a little bit temperamental.
Is it worth over 1,500?

PS would you know how to search archival e-mails pertaining to the group discussion of version 15?

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]]On Behalf Of
Automatic digest processor
Sent: Monday, July 30, 2007 12:00 AM
To: Recipients of SPSSX-L digests
Subject: SPSSX-L Digest - 28 Jul 2007 to 29 Jul 2007 (#2007-211)


There are 8 messages totalling 524 lines in this issue.

Topics of the day:

  1. Increasing decimal places for p-values
  2. MATCHing a date range to a date
  3. CTables and Significance in One Table (2)
  4. Multinomial Logistic Regression
  5. Return variable name as value in new variable
  6. Running a Factor Analysis with only the correlation matrix (2)

----------------------------------------------------------------------

Date:    Sat, 28 Jul 2007 22:08:27 -0600
From:    ViAnn Beadle <[hidden email]>
Subject: Re: Increasing decimal places for p-values

You can edit/activate the table by double-clicking on it and change the cell
format for the cell containing the p value but I'm not sure why you would
want to. Surely 3 decimal digits is enough to accept/reject?


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
[hidden email]
Sent: Saturday, July 28, 2007 9:13 PM
To: [hidden email]
Subject: Increasing decimal places for p-values

Does anyone know how (or if) one can increase the decimal places for
p-values
when you run a Chi-square test? The Asymp. Sig only goes to three and I
can't
find any options under Crosstab to change that. (How do you view very low p-
values?)
Thank you for your time and assistance.
Debra

--

------------------------------

Date:    Sun, 29 Jul 2007 03:28:12 -0700
From:    Albert-jan Roskam <[hidden email]>
Subject: Re: MATCHing a date range to a date

Hi Nico!

How do you know when two records from actual and
estimated values are corresponding records? Are they
sorted in such a way that record #1,2,3,n in Estimated
corresponds to record #1,2,3,n in Actual? Quite scary,
as it could really mess up things when you sort the
data.

I am going to assume you named your datasets
'estimated' and 'actual'.

dataset activate estimated.
compute casenum = $casenum.
dataset name estimated.

dataset activate actual.
compute casenum = $casenum.
dataset name actual.

match files
  / file = estimated
  / file = actual
  / by = casenum.
exe.

compute flag = 0.
if (date ge dt_frm and date le dt_to) flag = 1.
value labels flag
   0 'outside range'
   1 'within range'.
fre flag.

Please note that this is not tested and that this
hinges upon the 'casenum' assumption.

Cheers!!
Albert-Jan


--- Nico Munting <[hidden email]> wrote:

> At the moment I am facing a challenge for which I
> have not been able to
> find an answer yet. I am hoping that someone reading
> this list will be able
> to help me.
>
> I have a dataset with estimates and a dataset with
> actual values
> encountered. These datasets contain the following
> data:
>
> ESTIMATES
>
> dt_frm       dt_to         profile   estim
> 1-JAN-2004   31-DEC-2004         1    0.87
> 1-JAN-2004   31-DEC-2004         2    0.45
> 1-JAN-2005    5-JUN-2005         1    0.77
> 1-JAN-2005   30-JUN-2005         2    0.48
>
>
> ACTUAL_VALUES
>
> date          profile   actual
> 31-JAN-2004         1     0.95
> 30-JUN-2004         1     0.85
> 31-JAN-2004         2     0.40
> 30-JUN-2004         2     0.50
> 25-FEB-2005         1     0.80
> 31-MAR-2005         2     0.49
>
>
> The from and to dates in the datafile containing the
> estimates are unique
> on a given date and would not overlap in combination
> with a profile.
>
> In order to determine to what estimate a certain
> actual value should be
> compared with I wanted to do a match on profile with
> a condition that DATE
> should lie between the dates in DT_FRM and DT_TO. Is
> there a way in which I
> could write SPSS-code doing this type of join. I
> have been trying to come
> up with a way that could do the join automatically,
> but I have not been
> able to solve this problem.
>
> Any help would be appreciated!
>
> Regards,
>
> Nico Munting
>

Cheers!
Albert-Jan

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Did you know that 87.166253% of all statistics claim a precision of results that is not justified by the method employed? [HELMUT RICHTER]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


      ____________________________________________________________________________________
Luggage? GPS? Comic books?
Check out fitting gifts for grads at Yahoo! Search
http://search.yahoo.com/search?fr=oni_on_mail&p=graduation+gifts&cs=bz

------------------------------

Date:    Sun, 29 Jul 2007 14:09:04 -0400
From:    Henry Crawford <[hidden email]>
Subject: Re: CTables and Significance in One Table

I love the looks of this routine, and I must be doing something simple and
incorrect.  I run a great deal of python routines now every day in my spss
work.  However, this simple-looking use of the tables.py module
consistently gives me the same error in spss 14.02:

Traceback (most recent call last):
  File "<string>", line 12, in ?
  File "C:\Python24\lib\site-packages\tables.py", line 164, in mergeLatest
    raise ValueError, "Exactly two table numbers must be specified."
ValueError: Exactly two table numbers must be specified.

I don't know how to get past this, and achieve the much-sought-after
single table of statistical significance.

Help?

-Hank

------------------------------

Date:    Sun, 29 Jul 2007 12:12:19 -0600
From:    Beyhan Titiz <[hidden email]>
Subject: Re: Multinomial Logistic Regression

Hi all, thanks for the responses.
It looks like zero frequencies issue is not a big problem, but
quasi-complete separation is which apparently can be detected from low
likelihood ratios of each effect that contributes to the model. Yet
when I take out low categorical IVs, I still got the quasi-complete
separation warning.
Any more ideas for causes and how to remedy this?
thank you
Beyhan T. Maybach, PhD


On 7/29/07, Kathryn Gardner <[hidden email]> wrote:

> Hi,
>
> I don't know about multinominal reg as I haven't used it properly, but in
> ordinal reg a similar message about zero frequencies comes up because of the
> continuous predictors in the model. The continuous predictor/s produces a
> large number of cells (i.e., dependent variable levels by combinations of
> predictor variables) with zero frequencies. I've read that this is OK and
> normal, but large amounts of empty cells make model fit statistics
> unreliable, and they shouldn't therefore be used. I assume the same applies
> to multinominal reg.
>
> Kathryn
>
> > Date: Fri, 27 Jul 2007 13:33:57 -0400
> > From: [hidden email]
> > Subject: Multinomial Logistic Regression
> > To: [hidden email]
> >
> > Hi,
> > I am analyzing a dataset with 4 continuous and 14
> > categorical Ivs with 7 group levels as the
> > DV using MLR.
> >
> > Basically, multiple characteristics (continuous and categorical IVs) of
> > 7 tree Genera groups were recorded (taxonomically decided but
> > debatable Generas) to determine the set of characteristics that allows
> > for the best discrimination/classification between these Genera and if
> > the cases really fit in these Genera groups.
> >
> > In MLR, I am getting two errors; '85% of the cells have zero
> > frequency' and 'there is possibly a quasi-complete
> > separation in the data'. What could I do about 85% zero freguency and
> > and the quasi-complete seperation? There are no zeros in the cells so that
> > it doesnot result in odds ratio of infinity. What are the other reasons
> > for quasi-complete seperation?
> >
> > I would appreciate any suggestions to get this run going.
> > Sincerely
> >
> >
> > Beyhan Titiz Maybach, PhD
>
>
> ________________________________
> The future of MSN Messenger! Windows Live Messenger

------------------------------

Date:    Sun, 29 Jul 2007 15:01:09 -0500
From:    "Peck, Jon" <[hidden email]>
Subject: Re: CTables and Significance in One Table

If you can post your code, I'll take a look, although I will be =
traveling this week and have only intermittent email access.

The error message means that the function did not get the two table =
numbers that should be merged.

You can use the mergelatest routine if the output is at the end of the =
Viewer, or you can use one of the other routines that specifies specific =
item numbers to merge.

-Jon Peck


-----Original Message-----
From: SPSSX(r) Discussion on behalf of Henry Crawford
Sent: Sun 7/29/2007 1:09 PM
To: [hidden email]
Subject:      Re: [SPSSX-L] CTables and Significance in One Table
=20
I love the looks of this routine, and I must be doing something simple =
and
incorrect.  I run a great deal of python routines now every day in my =
spss
work.  However, this simple-looking use of the tables.py module
consistently gives me the same error in spss 14.02:

Traceback (most recent call last):
  File "<string>", line 12, in ?
  File "C:\Python24\lib\site-packages\tables.py", line 164, in =
mergeLatest
    raise ValueError, "Exactly two table numbers must be specified."
ValueError: Exactly two table numbers must be specified.

I don't know how to get past this, and achieve the much-sought-after
single table of statistical significance.

Help?

-Hank

------------------------------

Date:    Sun, 29 Jul 2007 10:08:28 +0200
From:    hillel vardi <[hidden email]>
Subject: Re: Return variable name as value in new variable

Shalom

here is a simple way of doing it if your variable list is of the form
var1 to var100 .

DATA LIST LIST /case_lbl (A4) var1 to var5 .
BEGIN DATA
"var1" . .09 .15 .10 .04
"var2" .09 . .01 .04 .17
"var3" .15 .01 . .06 .01
"var4" .10 .04 .06 . .05
"var5" .04 .17 .01 .05
END DATA .
compute         maxvarval=0 .
do repeat        var=var1 to var5/
                     num=1 to 5.
do if                var gt maxvarval .
compute          maxvarval=var .
compute          maxvarpos=num.
end if .
end repeat.
string              maxvarname(a6).
compute         maxvarname=concat('var',string(maxvarpos,f1)) .
list                  case_lbl  var1 to var5 maxvarval maxvarname .

If your variable names are not consecutive  you can copy the variables
name from the variable view and pest them into the do repeat command .

Hillel Vardi



Jon Bernard wrote:

> Hello SPSSers,
>
> I have a correlation matrix of 105 variables.  The values in the
> matrix are
> all positive, and the diagonals are missing.  Here are some sample data:
>
> DATA LIST LIST /case_lbl (A4) var1 to var5 .
> BEGIN DATA
> "var1" . .09 .15 .10 .04
> "var2" .09 . .01 .04 .17
> "var3" .15 .01 . .06 .01
> "var4" .10 .04 .06 . .05
> "var5" .04 .17 .01 .05
> END DATA .
>
>> From these data, I am trying to compute 2 new variables: rMax and
>> varMax.
> rMax is the maximum row value among the values in var1 to var5, and
> this is
> easy to compute with "COMPUTE rMax=MAX(var1 to var5)".  varMax is
> (hopefully) the variable name of the maximum row value, and this I cannot
> figure out how to compute.  The resulting data from a working solution
> would
> look like the sample data below:
>
> DATA LIST LIST /case_lbl (A4) var1 to var5 rMax * varMax (A4) .
> BEGIN DATA
> "var1" . .09 .15 .10 .04 .15 "var3"
> "var2" .09 . .01 .04 .17 .17 "var5"
> "var3" .15 .01 . .06 .01 .15 "var1"
> "var4" .10 .04 .06 . .05 .10 "var1"
> "var5" .04 .17 .01 .05 . .17 "var2"
> END DATA .
>
> Can anyone help with a reference or solution?  I should note here that
> while
> I have some experience with basic syntax, I have none with macros,
> scripts,
> or Python.
>
> Very many thanks in advance for your advice.  Please let me know if more
> information is necessary to propose a solution.
>
> -Jon
>

------------------------------

Date:    Sun, 29 Jul 2007 15:23:24 -0700
From:    klinas <[hidden email]>
Subject: Running a Factor Analysis with only the correlation matrix

Hello, I was wondering if it is possible to run a factor analysis if you just
have the correlation matrix and do not have access to the "raw" data.  Thank
you in advance!!
--
View this message in context: http://www.nabble.com/Running-a-Factor-Analysis-with-only-the-correlation-matrix-tf4167145.html#a11855788
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

------------------------------

Date:    Sun, 29 Jul 2007 19:41:26 -0400
From:    "Edgar F. Johns" <[hidden email]>
Subject: Re: Running a Factor Analysis with only the correlation matrix

"Hello, I was wondering if it is possible to run a factor analysis if you
just have the correlation matrix and do not have access to the "raw" data."


I did this a few weeks ago (with assistance from some list members). Below
is the syntax I used:

MATRIX DATA

VARIABLES=ROWTYPE_                      ECI1    ECI2    ECI3    ECI4    ECI5
ECI6    ECI7    ECI8    ECI9    ECI10   ECI11   ECI12   ECI13   ECI14
ECI15   ECI16   ECI17   ECI18   ECI19   ECI20   .
BEGIN DATA

MEAN    5.32    4.99    5.52    5.17    5.23    5.8     5.05    5.27    5.18
5.27    5.57    5.38    5.09    5.02    4.97    5.17    5.04    4.94    5.45
5.44
STDDEV  0.67    0.77    0.66    0.84    0.7     0.72    0.62    0.69    0.67
0.74    0.7     0.78    0.79    0.78    0.65    0.71    0.79    0.72    0.75
0.71
N       3391    3391    3391    3391    3391    3391    3391    3391    3391
3391    3391    3391    3391    3391    3391    3391    3391    3391    3391
3391
CORR    1

CORR    0.68    1

CORR    0.43    0.53    1

CORR    0.73    0.79    0.37    1

CORR    0.51    0.61    0.57    0.6     1

CORR    0.49    0.61    0.5     0.6     0.58    1

CORR    0.44    0.63    0.29    0.74    0.53    0.5     1

CORR    0.63    0.67    0.62    0.63    0.57    0.56    0.49    1

CORR    0.4     0.57    0.46    0.58    0.52    0.54    0.51    0.56    1

CORR    0.57    0.76    0.59    0.69    0.64    0.6     0.63    0.61    0.52
1

CORR    0.48    0.64    0.78    0.48    0.62    0.57    0.4     0.7     0.6
0.66    1

CORR    0.5     0.67    0.79    0.49    0.6     0.6     0.37    0.64    0.5
0.74    0.81    1

CORR    0.65    0.73    0.58    0.74    0.64    0.64    0.57    0.7     0.57
0.67    0.7     0.65    1

CORR    0.51    0.66    0.74    0.57    0.65    0.61    0.48    0.66    0.56
0.67    0.77    0.75    0.78    1

CORR    0.56    0.65    0.68    0.57    0.7     0.64    0.45    0.64    0.44
0.69    0.74    0.76    0.7     0.77    1

CORR    0.64    0.73    0.62    0.74    0.65    0.67    0.58    0.64    0.5
0.73    0.64    0.68    0.74    0.73    0.75    1

CORR    0.51    0.62    0.73    0.51    0.61    0.54    0.41    0.7     0.45
0.65    0.79    0.77    0.72    0.81    0.77    0.67    1

CORR    0.64    0.69    0.54    0.73    0.63    0.6     0.63    0.72    0.5
0.7     0.64    0.61    0.75    0.7     0.72    0.71    0.7     1

CORR    0.62    0.75    0.57    0.69    0.63    0.61    0.51    0.55    0.46
0.67    0.56    0.67    0.66    0.68    0.68    0.71    0.6     0.62    1

CORR    0.64    0.81    0.51    0.83    0.67    0.67    0.68    0.67    0.66
0.72    0.6     0.63    0.77    0.75    0.68    0.77    0.66    0.74    0.79
1
END DATA.



FACTOR
/MATRIX IN(COR=*)
/ANALYSIS  ECI1 ECI2    ECI3    ECI4    ECI5    ECI6    ECI7    ECI8    ECI9
ECI10   ECI11   ECI12   ECI13   ECI14   ECI15   ECI16   ECI17   ECI18
ECI19   ECI20
 /PRINT ROTATION
 / FORMAT SORT
/PLOT EIGEN ROTATION
  /CRITERIA FACTORS(2) ITERATE(25)
  /EXTRACTION paf
  /CRITERIA ITERATE(25)
 / ROTATION PROMAX(4)
 / METHOD=CORRELATION .


In doing so, I created the Correlation dataset according to the SPSS
instructions in the reference guide.

Hope this helps.
Edgar
---
Discover Technologies
42020 Koppernick Rd.
Suite 204
Canton, MI 48187
[phone]
[phone] fax
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
klinas
Sent: Sunday, July 29, 2007 6:23 PM
To: [hidden email]
Subject: Running a Factor Analysis with only the correlation matrix

you in advance!!
--
View this message in context:
http://www.nabble.com/Running-a-Factor-Analysis-with-only-the-correlation-ma
trix-tf4167145.html#a11855788
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

------------------------------

End of SPSSX-L Digest - 28 Jul 2007 to 29 Jul 2007 (#2007-211)
**************************************************************