Help with Multivariate, analyze demographics and response percentages

[Doc2Be]

Hi everyone. I am at the limits of my limited SPSS knowledge, and am in need of your know-how and experience.

I am trying to figure out how to analyze demographics and response percentages in a sample (non-normal).

I have 31 participants, answering 25 questions (outcome measures - numeric, scale), each on a scale of 1-5. I have four potential grouping vairables (numeric): Gender (nominal), age range (nominal), condition (nominal) and group (ordinal).

I'm trying to find a way to see what percentage of the respondents answered a 4 or above on each question and on average overall, and also break that down by grouping (i.e. how many women answered 4 or higher, overall, across the questions; how many in the 20-25 age range answered 4 or higher, etc.).

I can create another set of variables of averages for each respondent on their questions overall. But I dont know if its good to create what is essentially a bunch of excel formulas, when SPSS might be able to help me run these tests.

Any thoughts?  I really do appreciate your experience and insight.

[John F Hall]

With only 31 cases, percentages are a waste of time, even for the whole data
set.  I doubt if means will mean much either.

John F Hall (Mr)
[Retired academic survey researcher]

Email:   [hidden email]  
Website: www.surveyresearch.weebly.com
SPSS start page:  www.surveyresearch.weebly.com/1-survey-analysis-workshop



-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Doc2Be
Sent: 07 January 2015 12:46
To: [hidden email]
Subject: Help with Multivariate, analyze demographics and response
percentages

Hi everyone. I am at the limits of my limited SPSS knowledge, and am in need
of your know-how and experience.

I am trying to figure out how to analyze demographics and response
percentages in a sample (non-normal).

I have 31 participants, answering 25 questions (outcome measures - numeric,
scale), each on a scale of 1-5. I have four potential grouping vairables
(numeric): Gender (nominal), age range (nominal), condition (nominal) and
group (ordinal).

I'm trying to find a way to see what percentage of the respondents answered
a 4 or above on each question and on average overall, and also break that
down by grouping (i.e. how many women answered 4 or higher, overall, across
the questions; how many in the 20-25 age range answered 4 or higher, etc.).

I can create another set of variables of averages for each respondent on
their questions overall. But I dont know if its good to create what is
essentially a bunch of excel formulas, when SPSS might be able to help me
run these tests.

Any thoughts?  I really do appreciate your experience and insight.



--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Help-with-Multivariate-analyze
-demographics-and-response-percentages-tp5728363.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Doc2Be]

Yes, I completely agree. That said, this is about doing the analysis (from a purely "learning SPSS" standpoint) and not about the actual use of the analysis for anything practical or academic.

For the sake of my learning though, do you say that because of the low statistical power of the sample size?

[John F Hall]

In haste as my dinner is on the table.

Look at RECODE for grouping the scores, then CROSSTABS for comparing them
across demographics.

There are tutorials on this for absolute beginners on my site.


John F Hall (Mr)
[Retired academic survey researcher]

Email:   [hidden email]  
Website: www.surveyresearch.weebly.com  
SPSS start page:  www.surveyresearch.weebly.com/1-survey-analysis-workshop





-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Doc2Be
Sent: 07 January 2015 18:50
To: [hidden email]
Subject: Re: Help with Multivariate, analyze demographics and response
percentages

Yes, I completely agree. That said, this is about doing the analysis (from a
purely "learning SPSS" standpoint) and not about the actual use of the
analysis for anything practical or academic.

For the sake of my learning though, do you say that because of the low
statistical power of the sample size?



--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Help-with-Multivariate-analyze
-demographics-and-response-percentages-tp5728363p5728365.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Bruce Weaver]

Here's another similar approach (untested).  You'll have to fill in your own variable names, of course.

DO REPEAT old = q1 to q25 / new = High1 to High25.
+  COMPUTE new = old GE 4.
END REPEAT.
FORMATS High1 to High25 (F1).
CROSSTABS Gender age_range condition group BY High1 to High25 /
 CELLS = count row col.

John F Hall wrote

In haste as my dinner is on the table.

Look at RECODE for grouping the scores, then CROSSTABS for comparing them
across demographics.

There are tutorials on this for absolute beginners on my site.


John F Hall (Mr)
[Retired academic survey researcher]

Email:   [hidden email]   
Website: www.surveyresearch.weebly.com  
SPSS start page:  www.surveyresearch.weebly.com/1-survey-analysis-workshop





-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Doc2Be
Sent: 07 January 2015 18:50
To: [hidden email]
Subject: Re: Help with Multivariate, analyze demographics and response
percentages

Yes, I completely agree. That said, this is about doing the analysis (from a
purely "learning SPSS" standpoint) and not about the actual use of the
analysis for anything practical or academic.

For the sake of my learning though, do you say that because of the low
statistical power of the sample size?



--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Help-with-Multivariate-analyze
-demographics-and-response-percentages-tp5728363p5728365.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Doc2Be]

So, forgive me, because I have only been doing this for a couple of months, but I take that and paste it into the syntax editor (of course, putting the variable names in there),  is that what you are saying?

[Bruce Weaver]

Yes.

Doc2Be wrote

So, forgive me, because I have only been doing this for a couple of months, but I take that and paste it into the syntax editor (of course, putting the variable names in there),  is that what you are saying?

[John F Hall]

Bruce's suggestion creates 25 new variables (and is safer if you don't want
to permanently change your original 25 variables). Assuming your
demographics are all together in a block and q1 to q25 have no values
outside the range 0 - 5, three suggestions:

1:  Tabulate before recoding.
2:  Temporary recode (otherwise original values will be lost) then tabulate.
3:   Temporary recode (otherwise original values will be lost) then use
multiple response in dichotomous mode.  Produces a smaller table with
(recoded) value 4 only for q1 to q25

For demographics with many values, 1 and 2 above will create tables with
more cells than you have cases.

CROSSTABS gender to group by q1 to q25 /cel cou row.

TEMP.
RECODE q1 to q25 (5 = 4).
CROSSTABS gender to group by q1 to q25 /cel cou row.


TEMP.
RECODE q1 to q25 (5 = 4).
MULT RESPONSE groups = scales 'Scoring 4 or 5'
        (q1 to q25 (4))
        /variables = gender to group (1,20)
        /tables = gender to group by scales
        /cel cou row.

3 is a summary solution.  The range of 1,20 is a guess at the largest number
of categories, but you can specify the ranges separately eg:

        /variables = gender (1,2) age_range (1,6)
                condition (1,6) group (1,4)

If you haven't already done so, you should set the formats for all your
existing variables to (F1.0) as otherwise, if your output is set to values,
or values and labels, they will have two superfluous zeros in the tables.

I assume your data are a test for a larger study rather than an assignment
for assessment.  If so I really recommend you to have a look at Block 3
(Analysing two variables) on my site:

http://surveyresearch.weebly.com/block-3-analysing-two-variables-and-sometim
es-three.html

Since you're a newbie to SPSS (and possibly to questionnaire surveys) it
won't do you any harm to look at blocks 1 and 2 as well:

http://surveyresearch.weebly.com/block-1-from-questionnaire-to-spss-saved-fi
le.html

http://surveyresearch.weebly.com/block-2-analysing-one-variable.html

Teach-yourself SPSS and look, no equations!  (at least not until you get to
Block 4 section 2.1)

As a rule of thumb, percentages should only be reported when the base is 40
or more cases, the reason being that, with 40 cases, if you move a case from
one category to another, it makes a net difference of 5 percentage points
(-2.5% in the source cell and +2.5% in the target cell).

John F Hall (Mr)
[Retired academic survey researcher]

Email:   [hidden email]  
Website: www.surveyresearch.weebly.com  
SPSS start page:  www.surveyresearch.weebly.com/1-survey-analysis-workshop




-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Bruce Weaver
Sent: 07 January 2015 20:04
To: [hidden email]
Subject: Re: Help with Multivariate, analyze demographics and response
percentages

Here's another similar approach (untested).  You'll have to fill in your own
variable names, of course.

DO REPEAT old = q1 to q25 / new = High1 to High25.
+  COMPUTE new = old GE 4.
END REPEAT.
FORMATS High1 to High25 (F1).
CROSSTABS Gender age_range condition group BY High1 to High25 /  CELLS =
count row col.



John F Hall wrote

> In haste as my dinner is on the table.
>
> Look at RECODE for grouping the scores, then CROSSTABS for comparing
> them across demographics.
>
> There are tutorials on this for absolute beginners on my site.
>
>
> John F Hall (Mr)
> [Retired academic survey researcher]
>
> Email:  

> johnfhall@

>    
> Website: www.surveyresearch.weebly.com SPSS start page:  
> www.surveyresearch.weebly.com/1-survey-analysis-workshop
>
>
>
>
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:

> SPSSX-L@.UGA

> ] On Behalf Of
> Doc2Be
> Sent: 07 January 2015 18:50
> To:

> SPSSX-L@.UGA

> Subject: Re: Help with Multivariate, analyze demographics and response
> percentages
>
> Yes, I completely agree. That said, this is about doing the analysis
> (from a purely "learning SPSS" standpoint) and not about the actual
> use of the analysis for anything practical or academic.
>
> For the sake of my learning though, do you say that because of the low
> statistical power of the sample size?
>
>
>
> --
> View this message in context:
> http://spssx-discussion.1045642.n5.nabble.com/Help-with-Multivariate-a
> nalyze -demographics-and-response-percentages-tp5728363p5728365.html
> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>
> =====================
> To manage your subscription to SPSSX-L, send a message to

> LISTSERV@.UGA

>  (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

> LISTSERV@.UGA

>  (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





-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Help-with-Multivariate-analyze
-demographics-and-response-percentages-tp5728363p5728371.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

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

[Doc2Be]

Bruce and John,

Thank you both, so very much. I am processing your answers. Already, using the recode, I am getting mostly what I am looking for.

I am going to keep plugging away... thank you.

John, you mentioned, in telling me about not reporting percentages for under 40 cases, that "it makes a net difference of 5 percentage points".  Will you help me see with your experience? Is 5 percentage points too much?

As I get more into doctoral level statistics, I am finding a difficulty in learning what spread of numbers is large and what is not (and I understand that it depends on the subject at hand). I might look at a number that is .368 and a number that is .395, and not think much, because its only a .027 difference, but then I keep reading and the researcher will say, "...is all the way up to .395 - a very large difference" and I am stymied.

I know that's a bit off-thread for our topic.  But please, help me understand the 5 percentage points being a big deal.

Thanks.