SPSSX Discussion

Comparing Selected Cases with All Cases

Classic

List

Threaded

4 messages Options

Dybdahl, Alf-Christian

Comparing Selected Cases with All Cases

Dear all,

Happy New Year!
I am using SPSS 15.0 for Windows.

I have a file with responses from a survey from different countries and
departments. I want to run simple statistics on the results where I e.g.
create a report that shows the results for Germany compared with the
total scores.

I am pretty new to SPSS and would greatly appreciate any help on how to
accomplish this.

Thanks for your time.

Best regards,
Alf-Christian Dybdahl
Laerdal Medical

====================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-2

Re: Comparing Selected Cases with All Cases

You most likely want to compare Germany to the other countries rather
than having Germany on both sides of the comparison.
First create a variable the represents that comparison of interest.
Something like this.

compute mygroups = country eq 'Germany'.
value labels mygroups 1 'Germany' 0 'other countries'.

A lot depends on your sample design. Are there already case weights in
the data? etc. These are necessary to weight the cases,

My superficial guess would be that you would want to use the COMPLEX
SAMPLES procedures.

A lot depends on the questions you want to ask of the data and the
levels of measurement of the data.

You might do simple comparisons of percentages, t-tests, contingency
tables, or more sophisticated models of the data such as categorical
regression, discriminant function analysis, etc.

Art Kendall
Social Research Consultants

Dybdahl, Alf-Christian wrote:

> Dear all,
>
> Happy New Year!
> I am using SPSS 15.0 for Windows.
>
> I have a file with responses from a survey from different countries and
> departments. I want to run simple statistics on the results where I e.g.
> create a report that shows the results for Germany compared with the
> total scores.
>
> I am pretty new to SPSS and would greatly appreciate any help on how to
> accomplish this.
>
> Thanks for your time.
>
> Best regards,
> Alf-Christian Dybdahl
> Laerdal Medical
>
>
> ===================
> 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
>
>
>

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

Dybdahl, Alf-Christian

Re: Comparing Selected Cases with All Cases

Hi!

Thanks for the reply!

I only want to run 2 simply analysises for now:

1. Case Summaries (showing the mean score - all questions are scores
from 1 to 6)
2. Frequency analysis (graph).

Thank you
Alf.Christian Dybdahl

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Art Kendall
Sent: 3. januar 2008 14:43
To: [hidden email]
Subject: Re: Comparing Selected Cases with All Cases

You most likely want to compare Germany to the other countries rather
than having Germany on both sides of the comparison.
First create a variable the represents that comparison of interest.
Something like this.

compute mygroups = country eq 'Germany'.
value labels mygroups 1 'Germany' 0 'other countries'.

A lot depends on your sample design. Are there already case weights in
the data? etc. These are necessary to weight the cases,

My superficial guess would be that you would want to use the COMPLEX
SAMPLES procedures.

A lot depends on the questions you want to ask of the data and the
levels of measurement of the data.

You might do simple comparisons of percentages, t-tests, contingency
tables, or more sophisticated models of the data such as categorical
regression, discriminant function analysis, etc.

Art Kendall
Social Research Consultants

Dybdahl, Alf-Christian wrote:
> Dear all,
>
> Happy New Year!
> I am using SPSS 15.0 for Windows.
>
> I have a file with responses from a survey from different countries
> and departments. I want to run simple statistics on the results where
I e.g.

> create a report that shows the results for Germany compared with the
> total scores.
>
> I am pretty new to SPSS and would greatly appreciate any help on how
> to accomplish this.
>
> Thanks for your time.
>
> Best regards,
> Alf-Christian Dybdahl
> Laerdal Medical
>
>
> ===================
> 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

ViAnn Beadle

Re: Comparing Selected Cases with All Cases

One way to do this is to create copies of the score questions except that
the value is missing for the copy unless the case is one for Germany. It's
fairly easy to get the case summary for both variables in a single table but
a bit harder to get a nice clustered bar chart with clusters defined by
Germany vs. All. This syntax assumes that you have one variable identifying
the country and one score variable.

1. Case summaries. In this instance I've used CTABLES but you can also get
summaries from DESCRIPTIVES and SUMMARIZE:

if (country eq 'Germany')gscore=score.
variable labels gscore 'Score for Germany'.
CTABLES
/VLABELS VARIABLES=gscore score DISPLAY=DEFAULT
/TABLE gscore [MEAN] + score [MEAN].

2a. Frequencies side by side.
GGRAPH
/GRAPHDATASET NAME="graphdataset" VARIABLES=score[LEVEL=ORDINAL] gscore
[LEVEL=ORDINAL]
MISSING=VARIABLEWISE
REPORTMISSING=NO
/GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
SOURCE: s=userSource(id("graphdataset"))
DATA: score=col(source(s), name("score"), unit.category())
DATA: gscore=col(source(s), name("gscore"), unit.category())
GUIDE: axis(dim(1), label("score"))
GUIDE: axis(dim(2), label("Frequency"))
SCALE: linear(dim(2), include(0))
ELEMENT: interval(position(summary.count(score/"Overall
Score"+gscore/"Germany Score"
)), shape.interior(shape.square))
END GPL.

2b. Clustered frequencies. This is best done by first restructuring so that
you get two rows. The first row contains the value of the score variable in
a new variable and the second row contains the value of the gscore variable
in that same variable (which will be missing for all non-German rows). An
index variable is then used to do the clustering. Note that VARSTOCASES will
overwrite your active data file so make sure it is saved first before
running. This syntax assumes that there is no unique case id:

VARSTOCASES
/MAKE newscore FROM score gscore
/INDEX=index "Germany vs. All"(2)
/KEEP=country
/NULL=KEEP.
value labels index 1 "All" 2 "Germany".

To extend this to multiple variables use the Restructure wizard to create
multiple variables corresponding to the score/gscore pair for each.

After restructuring you have a simple clustered bar chart that could be
generated using Chart Builder:

GGRAPH
/GRAPHDATASET NAME="graphdataset" VARIABLES=newscore[LEVEL=ORDINAL]
COUNT()[name="COUNT"] index
MISSING=LISTWISE REPORTMISSING=NO
/GRAPHSPEC SOURCE=INLINE.
BEGIN GPL
SOURCE: s=userSource(id("graphdataset"))
DATA: newscore=col(source(s), name("newscore"), unit.category())
DATA: COUNT=col(source(s), name("COUNT"))
DATA: index=col(source(s), name("index"), unit.category())
COORD: rect(dim(1,2), cluster(3,0))
GUIDE: axis(dim(3), label("newscore"))
GUIDE: axis(dim(2), label("Count"))
GUIDE: legend(aesthetic(aesthetic.color.interior), label("Germany vs.
All"))
SCALE: linear(dim(2), include(0))
ELEMENT: interval(position(index*COUNT*newscore), color.interior(index),
shape.interior(shape.square))
END GPL.

You'll need a separate GGRAPH command for each score variable.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Dybdahl, Alf-Christian
Sent: Thursday, January 03, 2008 7:36 AM
To: [hidden email]
Subject: Re: Comparing Selected Cases with All Cases

Hi!

Thanks for the reply!

I only want to run 2 simply analysises for now:

1. Case Summaries (showing the mean score - all questions are scores
from 1 to 6)
2. Frequency analysis (graph).

Thank you
Alf.Christian Dybdahl

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Art Kendall
Sent: 3. januar 2008 14:43
To: [hidden email]
Subject: Re: Comparing Selected Cases with All Cases

You most likely want to compare Germany to the other countries rather
than having Germany on both sides of the comparison.
First create a variable the represents that comparison of interest.
Something like this.

compute mygroups = country eq 'Germany'.
value labels mygroups 1 'Germany' 0 'other countries'.

A lot depends on your sample design. Are there already case weights in
the data? etc. These are necessary to weight the cases,

My superficial guess would be that you would want to use the COMPLEX
SAMPLES procedures.

A lot depends on the questions you want to ask of the data and the
levels of measurement of the data.

You might do simple comparisons of percentages, t-tests, contingency
tables, or more sophisticated models of the data such as categorical
regression, discriminant function analysis, etc.

Art Kendall
Social Research Consultants

Dybdahl, Alf-Christian wrote:
> Dear all,
>
> Happy New Year!
> I am using SPSS 15.0 for Windows.
>
> I have a file with responses from a survey from different countries
> and departments. I want to run simple statistics on the results where
I e.g.

> list of commands to manage subscriptions, send the command INFO
> REFCARD
>
>
>

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

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

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