[openbeos-cdt] Re: Border for resizing (was: Re: Areas of interest for CDT)

  • From: Eddy Groen <eddyspeeder@xxxxxxxxx>
  • To: openbeos-cdt@xxxxxxxxxxxxx
  • Date: Fri, 13 Nov 2009 10:32:22 +0100

> That kind of thing would really be nice, but I doubt the results would
> be consistent, when not every single experiment is carefully set up and
> controlled. It may turn out to be only practical for very few, well
> defined tasks. And implementing those is a bitch, as Eddy could attest,
> I imagine.

In the most agreeing way possible, in real life I would sighingly say: "Uh,

That's why R&D is so "expensive"; it takes a lot of time and effort (thus,
money) and the short-term "gains" are rather small. And as Humdinger pointed
out, it needs to be controlled, meaning that to exclude confounding factors
you need to get people in the same place, using the same computer, doing the
same task.

The trick is now to come up with a way to make well-founded claims about
usability, as indeed we lack the physical and monetary resources to do this
"the pro way". I think there is a way.

Let's take the question Humdinger asks, meaning we stay on the "border for
resizing" topic. This can actually be measured pretty well using an online
questionnaire. Yes, questionnaires need to be carefully set up as well, but
we can still ask people about their self-reported use of window borders for
resizing. Self-reported measures are quite accurate, at least enough to
answer Humdinger's question with at least some certainty. For real simple
questions, a poll could indeed be used. Perhaps it would even be a good idea
to add a poll to the Haiku Web site. There are, however a few limitations to
a poll, which I think a periodic questionnaire could intercept:

1. A poll must have short questions and short answers. A poll like "How
often do you use the window borders to resize a window?" [always, mostly,
sometimes, never] does not adequately measure what we want to know (because
the question is open for multiple interpretations and the answer doesn't
really sum up the respondent's actual behavior). For instance, you can't
know from the answers when a user would sometimes do it by using the
borders, and at other times by the window corners or the fullscreen button.
You could, for instance give people a picture of a window with a text field
near the fullscreen button, one near the right side window corner, one near
the bottom side window corner, and one near the bottom right corner. Then
ask the question: "Please fill out, on average, what percentage of the time
you use any of these window features to resize the window." (Another field
should sum up the percentages and give an error when the value is not 100%.)

2. A poll can only be held once in so many time. Say, you put a poll up for
two weeks, then you can do 26 polls a year. It would actually be much more
useful to gather a steady database of a few hundred respondents of Haiku
users & developers who would like to fill out an online questionnaire, say,
once every two to three months. This will allow us to tap deeper into
questions such as window resizing behavior, while at the same time giving us
answers on a whole bunch of other questions that were raised in the two to
three months prior.


Other related posts: