Forum Poll Questions
Michael Behrens
michael.behrens at gmail.com
Tue Feb 19 22:19:04 UTC 2008
On Feb 19, 2008 5:04 PM, Onno Benschop <onno at itmaze.com.au> wrote:
> >
> > You don't need to talk about products. Just the architecture would be
> > interesting. The Sun servers in question are sparc. so the question may
> > be - i386, amd64, or sparc?
> >
> > dave
> >
> I suppose that it depends on what kind of data you're trying to capture.
>
i386 / amd64 == x86 in the overall sense of things, yes there are
differences
but I think if we create too many options, we will turn people off.
As for sparc, yes, there are certainly sparc users, in fact there are
ppc users (I just installed -server ppc myself on a test machine).
I think the best way is to have the final item in that list be
* Other (Please Specify)
That is, the original question was in a round-about kind of way asking
> about brands and presumably coverage within those brands to see if
> something falls out of the bottom, that is perhaps we'll see something
> like: "We know that 50% of our servers are run on HP hardware."
>
> The path Dave is taking is one of CPU/architecture.
>
> They're two different kinds of questions.
>
Agreed, they are two very different questions and while we are
likely to be addressing a technical audience with the survey, we cannot
garrauntee that, therefore it should be, IMO as general as possible, which
means sticking to brands.
| I should also point out that more questions isn't always better. It
> might be useful to look at other such surveys so we can compare results
> between our survey and those conducted by others.
>
Yes, This survey is again, my opinion, to give us a baseline on the
community.
>From there we can branch out and apply what we have learned, or correlate
it to other surveys.
>
> Finally, having conducted several ICT surveys myself in the past, it
> would in my opinion be beneficial for the survey outcome, to contact a
> statistician before you send out the survey to see if you missed any
> design issues.
>
Do we have an stats people on the team? or do we have the resources
to have one involved? I think those are two primary issues which we would
face in order to do so. I do not disagree with the idea however.
Michael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-server/attachments/20080219/222acf6f/attachment.html>
More information about the ubuntu-server
mailing list