Gnome3 survey results

[creeping death]

Some time back...i remember i had seen a guy asking the members to participate in gnome3 live survey...most members were skeptical about the authenticity of that survey on account of the fact that the survey was not hosted on the gnome site...

i dont know how many of you participated...but i got this e-mail announcing the results of the survey...just thought i ll share

interesting statistics...read on

http://live.gnome.org/UsabilityProje...eneralResearch

[pwca]

The sample size was too small.

[pete_1967]

Originally Posted by pwca

The sample size was too small.

Doesn't make the slightest difference when questions made it skewed from the start. Scary thing is, they might actually think the data is worth something.

Only accurate data they got from this so called survey was the open feedback - but that of course is unquantifiable.

[pwca]

Originally Posted by pete_1967

Doesn't make the slightest difference when questions made it skewed from the start. Scary thing is, they might actually think the data is worth something.

Only accurate data they got from this so called survey was the open feedback - but that of course is unquantifiable.

Yes, sample size does make a difference irrespective of the questions being asked. Although I agree with your assertion concerning the questions being flawed.

[pete_1967]

Originally Posted by pwca

Yes, sample size does make a difference irrespective of the questions being asked.

Well, yes, of course it does. What I meant that when the data is useless and worthless it should be immediately discarded anyway.

[dmyersturnbull]

>The sample size was too small.
The sample size is fine.
For a 95% confidence level, the sample size of 1000 gives an interval of 3.1 for 20,000,000 Gnome users (rough estimate which hardly matters) for questions where each answer is equally likely. That's high enough for any reasonable research.

But how was this sample chosen? I'm more skeptical about that.

[creeping death]

Originally Posted by dmyersturnbull

>The sample size was too small.
The sample size is fine.
For a 95% confidence level, the sample size of 1000 gives an interval of 3.1 for 20,000,000 Gnome users (rough estimate which hardly matters) for questions where each answer is equally likely. That's high enough for any reasonable research.

But how was this sample chosen? I'm more skeptical about that.

well, there was this guy who posted a thread, inviting members of this forum to participate in his survey...it was more like an open survey...everyone was invited to participate...the gnome guys probably went around to every major disto support forum and asked people to participate...and i think every participants voice was reflected in the survey...

i was expecting many more people would participate...i too felt that the sample was small...but i m no statistician so i cant comment on your explanation...i guess you are right...

i think they would take this survey into consideration, considering the fact that the results are published at gnome website...some of the conclusions they drew were useful i think...

for example:

People use blogs but most of them don't have websites. Why not make Gnome 3.0 integrate that properly in the form of auto-blogger interface based on their actions(activities)?

thats a nice thought...

People need a better (more intuitive) way to locate files on their computers.

i d love to see what they come up for handling this requirement...

[pete_1967]

Originally Posted by dmyersturnbull

But how was this sample chosen? I'm more skeptical about that.

They don't actually know:

I was surprised by the results of the workspace question, and it makes me wonder ... could we get a list of mailing lists and forums that were used to get those people to respond? I suspect that we may have captured just a segment of our users.

For example I sent out the invite to the Gnome-Shell mailing list.


Perhaps we may need to create a similar survey and explicitly look for particular user types (like non-technical people)?

Unless this is the definite answer:

It was posted on Ubuntu/Fedora forums and that's where unexpected
results came from but it's good because not everybody should know what a
workspace is and some seem to not care at all.

[robert-e]

This qualifier will tend to skew the result
"People use blogs but most of them don't have websites." It should not be there. Instead, just ask the question thus:

"Would you use an AutoBlog interface if available on the Desktop" Actually, I am leary of the "if available on the Desktop" portion.

Writing good poll questions is very difficult; It takes a lot of time and effort to get statistically sound results.

Personally, I view most of the polls showing up in user-groups and forums with a jaundiced eye; most of them are posted to get the result wanted by the "postee". To put it bluntly, they often have an axe to grind.

Regards,
Bob