Excluding responses which are not valid from frequency distribution analysis

[t]

We have some survey results which contain answers to questions like
"How satisfied are you with Parking Services", "How satisfied are you
with Payroll Services", "How satisfied are you with Reimbursement
Services". There are 30 such questions.

The responses are around 1000. Some respondents might not answer the
question "How satisfied are you with Parking Services" so there is a
blank in MS-Excel spreadsheet. Some may respond with a answer like "No
Opinion" or "Not Applicable"

I am using IBM SPSS 21. I understanding going to Analyze(option in the
menu)->Descriptive Statistics->Frequencies, I can get the percentage
distribution.

I am trying to get the frequency distribution of responses (i.e. how
many percentage of people are satisfied with Parking Services. So, if
out of 1000 respondents, 50 responses were blank or "No Opinion" or "Not
Applicable", 650 were "Very Satisfied", 50 were "Satisfied", 250 were
"Dissatisfied" it means 700/950 were number of people who were
satisfied(Adding people who are "Very Satisfied" or "Satisfied") which
is around 74%

How can I filter cases which have no response or have "No Opinion" so
that I can get the right number (700/950 or 74%)?

I realize I can use the "Select Cases" option from the "Data" dropdown
menu. Use the "If condition is satisfied" option and then click the "if"
box. Then select a variable and set it to <> missing value ("" or
"No Opinion" or "Not Applicable") then & the
next variable set to <> missing & the next variable, and so on. This can
give a list of all cases that don't have a missing value.

But, I don't want to filter like that. As a respondent John has no
opinion for Parking services, but can have views about Payroll services.
So, I don't want to filter all his responses.

Any suggestions would be appreciated.

Thanks

[Rich Ulrich]

see MISSING VALUES.

It sounds like you are starting with test-strings as responses,
and that is not convenient for later analyses.

You need to use RECODE or AUTORECODE to create a numeric
variable, and put all the NA responses into one or several categories,
and declare those categories as MISSING. Then FREQ will show
the percentages in both ways, % of total and % of Valid.

If you use AUTORECODE, it gives you values in alphabetical
order, which is likely to be inconvenient, too, so you would
want a further step of creating a variable that puts the good
responses as properly ordered to 1-5, or whatever.

--
Rich Ulrich

[t]

Thanks Rich,

I declared those categories as missing.

Some of the responses are "Highly Satisfied" and some are "Satisfied"
How do I get the percentages to be added? I mean for a question, if
percentage of people who chose "Highly Satisfied" is 60% and people who
chose "Satisfied" is 10%, is there a way, I can code SPSS to total them
so that it comes as 70%?

Thanks for your time and advice.

[David Marso]

See RECODE command!

[Rich Ulrich]

On Thu, 01 Aug 2013 20:57:35 -0400, t <t...@t.com> wrote:

[snip, previous]

>
>Thanks Rich,
>
>I declared those categories as missing.
>
>Some of the responses are "Highly Satisfied" and some are "Satisfied"
>How do I get the percentages to be added? I mean for a question, if
>percentage of people who chose "Highly Satisfied" is 60% and people who
>chose "Satisfied" is 10%, is there a way, I can code SPSS to total them
>so that it comes as 70%?
>
>Thanks for your time and advice.

Same basic logic as coding several MISSING to the same
numeric value -

RECODE v5point(1,2,3=1)(4,5=2) /INTO v2point.

Or maybe there's not a / before INTO.

Googling on SPSS RECODE INTO gets some pages
with information on using the GUI, which is a more
awkward and tedious way to create the same lines.
(Always paste your syntax and save as permanent
documentation.)

--
Rich Ulrich

[t]

Thanks, yes it worked. Any SPSS 101 tutorials you would recommend?

[t]

Thanks, yes it worked. Any SPSS 101 tutorials you would recommend?

I am going through the Help which comes with SPSS.

[Rich Ulrich]

On Fri, 02 Aug 2013 15:26:49 -0400, t <t...@t.com> wrote:

...snip, previous

>>
>Thanks, yes it worked. Any SPSS 101 tutorials you would recommend?
>I am going through the Help which comes with SPSS.

I haven't used any of the tutorial sites but the general one
most often mentioned in the past has been the one at UCLA.

See Bruce Weaver's page for a lot of other locations -
https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/spss/



--
Rich Ulrich

[t]

Thanks Rich,

I appreciate your assistance and time.

[t]

On 8/4/2013 1:31 AM, Rich Ulrich wrote:

Thanks Rich,

I appreciate your assistance and time.

P.S. I don't know why this did not get posted so sending it again.