Re: Multi response sets in PSPP

[Alan Mead]

I recently added a small feature to PSPP, and I think if someone had
money and wanted to fund a feature, it might be possible to do so.

For my small feature, I had to complete some paperwork to assign my
copyright to FSF. Then I added the feature to the code and tested it.
Then I modified the documentation.  Finally, I delivered a patch for the
source code and the documentation.  John Darrington and/or Benn Pfaff
turned my test information into actual tests.

Modifying PSPP to perform some function and modifying PSPPIRE to
configure that function are two different (independent) steps. My change
hasn't been added to PSPPIRE (the GUI dialog boxes) and some users would
consider a feature not to exist if it's not accessible through menus, so
you should probably add that as a last step.

So, if I were paying someone (like a freelancer or student), I'd start
by modifying the documentation and writing test cases. The documentation
would be for their use and the test cases would determine that they'd
done the work correctly.

It seems like you'd need to coordinate/negotiate with either Ben Pfaff
or the FSF to ensure that the person paying for the work will own the
copyright and be able to transfer it to the FSF (and, of course, it will
be necessary to do so if you wanted the feature added to PSPP). I assume
this isn't insurmountable.

Also, for technical reasons, it might be easier if the developer and
whomever was paying for the work were using PSPP on Linux. It's harder
to generate Windows binaries (but maybe the freelancer could manage
this) and it's relatively easy on Linux to apply patches and compile
PSPP to test the changes. I'm using CentOS 6 and everything works great
except I cannot compile PSPPIRE (due to limitations in CentOS 6 versions
of some libraries).  I think John Darrington uses Debian or Ubuntu.  I
think CentOS/RHEL 7 would also work better.

I'm sure CTABLES would be a lot of work (not a small feature), but the
actual changes to the PSPP source code might fit into a modest budget.
You'd have to get a price for an inexpensive freelance coder.

-Alan


On 11/6/2015 12:00 PM, Robert Lutener wrote:
> There is no interest from the developers to do this.  A while back someone 
> even offered to put down $500 of their own money and others volunteered a few 
> hundred each on top to bring this about but it was stated that there is no 
> interest from the developers to implement this functionality.  Which is kind 
> of weird because CTABLES is the key reason why the Canadian government uses 
> SPSS at CRA and a few other departments.  For their stats stuff they 
> typically use R or Stata.
>
> ----- Original Message -----
> From: "Matthias Faeth" <address@hidden>
> To: "robertl" <address@hidden>
> Cc: "ftr" <address@hidden>, "pspp-users" <address@hidden>
> Sent: Friday, November 6, 2015 9:32:53 AM
> Subject: Re: Multi response sets in PSPP
>
> I strongly support Robert's view. However, the maintainers seem to be more
> interested in statistics than plain old tabulation. Has anyone an idea how
> to implement even a stripped down version of CTABLES?
>
> Matthias Fäth
> Im Mediapark 12
> 50670 Köln
> t: 0221-2907973
> m: 0171-9832175
> e: address@hidden
>
> 2015-11-06 18:06 GMT+01:00 Robert Lutener <address@hidden>:
>
>> CTABLES is probably the most valuable feature of SPSS and is used by the
>> vast majority of people that buy the SPSS license.  It is a real shame that
>> PSPP doesn't have CTABLES as this would be a major factor in more
>> widespread use of this open source solution.
>>
>> ----- Original Message -----
>> From: "ftr" <address@hidden>
>> To: "pspp-users" <address@hidden>
>> Sent: Friday, November 6, 2015 5:56:23 AM
>> Subject: Re: Multi response sets in PSPP
>>
>> On 06/11/2015 11:24, Michał Dubrawski wrote:
>>
>>
>>
>> Hi,
>>
>> If the information here is up to date then we have MRSETS command
>> implemented in PSPP
>> http://www.gnu.org/software/pspp/manual/pspp.html#MRSETS
>>
>> Is there any way to display these MRSETS in the PSPP output - frequences,
>> percentages or in corss tabs? I'm only using thses MRSETS in SPSS in
>> CTables. I am aware that CTables are not implemented in PSPP and since it
>> is very complicated won't rather be implemented in the nearest future.
>>
>> If it is not possible to display multi response, maybe implementation of
>> MULT RESPONSE command would be a solution? Is it possoble to implement it?
>>
>> It is has an option to display frequences, percentages and even cross
>> tabulation. I think that it would be very useful for users, what do you
>> think?
>>
>> Thanks again to all the developers and people involved with PSPP project
>> for all your work.
>>
>> kind regards,
>>
>> Michał Dubrawski
>>
>> Hi Michal,
>>
>> survey analyst NEED methods to study multiple-response variables. So I
>> strongly support the idea to implement a Mult Response command.
>>
>> The help text ends with this cryptic sentence:
>>
>>
>> Otherwise, multiple response sets are currently used only by third party
>> software.
>> Dear maintainers, would you please be so kind to explain which third party
>> software is meant ?
>>
>> TIA
>> -ftr
>>
>>
>> _______________________________________________
>> Pspp-users mailing list
>> address@hidden
>> https://lists.gnu.org/mailman/listinfo/pspp-users
>>
>> _______________________________________________
>> Pspp-users mailing list
>> address@hidden
>> https://lists.gnu.org/mailman/listinfo/pspp-users
>>
> _______________________________________________
> Pspp-users mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/pspp-users

-- 

Alan D. Mead, Ph.D.
President, Talent Algorithms Inc.

science + technology = better workers

+815.588.3846 (Office)
+267.334.4143 (Mobile)

http://www.alanmead.org

Announcing the Journal of Computerized Adaptive Testing (JCAT), a
peer-reviewed electronic journal designed to advance the science and
practice of computerized adaptive testing: http://www.iacat.org/jcat