Problems with split file

Bjørn

unread,

Nov 24, 2010, 9:56:49 AM11/24/10

to

I wanna compare two countries and use my dependent variable (fear of
terror in the next 12 months) and see if there is any sig. with my
independet variables (age, gender, years of fulltime education).
I have dummycoded my DV into 1=yes 0=no, and also dummycoded my two
countries and my IV.

This is what I do: data-split file-put in my DV (dummyterror) and pres
OK. Then I go to analyze-regression-linear. Insert my DV
(dummyterrror) in the right box and my IV in the block 1 box and press
"ok"

but I don`t get the tables I want, instead I get errors like

"Model Summary(a)

a. There are no valid cases in one or more split files. Statistics
cannot be computed."

Hope some of you understand my problem, and can help

Bjorn
Norwegian Student
SPSS18

Bruce Weaver

unread,

Nov 24, 2010, 10:18:29 AM11/24/10

to

Bjorn, if you want to compare the two countries, you need the data for
both of them in a single model. So do not use SPLIT FILE.

Also, if your dependent variable is dichotomous (Yes/No), logistic
regression would be a more natural choice than linear regression. Was
it dichotomous originally? Or did you convert something else into a
dichotomy?

HTH.

--
Bruce Weaver
bwe...@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/Home
"When all else fails, RTFM."

Bjørn

unread,

Nov 24, 2010, 10:59:33 AM11/24/10

to

> bwea...@lakeheadu.cahttp://sites.google.com/a/lakeheadu.ca/bweaver/Home

> "When all else fails, RTFM."

Hi Bruce, I know, but since this is a term paper my professor adviced
us to do split file first, and then combine
My DV had 4 values from very likely to not likely at all, and I coded
it to yes/no (dichotmous)

Bruce Weaver

unread,

Nov 24, 2010, 1:21:40 PM11/24/10

to

With the original outcome variable, ordinal logistic regression would
be another possibility (via the PLUM procedure).

What is your sample size?

--
Bruce Weaver
bwe...@lakeheadu.ca

Kylie

unread,

Nov 24, 2010, 6:21:46 PM11/24/10

to

Hi Bjorn,

If you want to run a separate analysis for each country, then you
should be inserting COUNTRY as the Split File variable, not
DUMMYTERROR.

Cheers,
Kylie.

Bjørn

unread,

Nov 25, 2010, 9:36:38 AM11/25/10

to

Thanks Kylie, I got my table for one country, but how do I get both
countries in the same table?

Message has been deleted

peter m sopp

unread,

Nov 25, 2010, 10:08:06 AM11/25/10

to

Am 25.11.2010 15:38, schrieb Bjørn:
> On 25 Nov, 00:21, Kylie <kylie.la...@gmail.com> wrote:

> Thanks, Kylie...I got it right with my table, but how do I get two
> countries in the same table ?
SPLIT FILE LAYERED BY country? (I never used it with regression, only
with FREQUENCY or MEANS ... - maybe that's what your looking for ... or
your prof)
Peter
Peter

Bjørn

unread,

Nov 25, 2010, 10:20:26 AM11/25/10

to

No, not with split file but in a combinded model. I wanna have one
model with both countries in, testing for sig. with my DV and my IV
but don`t know how to do it !

peter m sopp

unread,

Nov 25, 2010, 11:28:12 AM11/25/10

to

Dummy? 1 = Country 1 und 0 = Country 2, so term country = 0 wenn country
2 (0*b) and 1*b wenn Country 1 - just one variable. That's the point?
Peter