[Usability] Re: Decision: instant apply window buttons

[John Kodis <kodis jagunet com>]

On Sun, Jan 20, 2002 at 05:20:13PM -0800, Seth Nickell wrote:

> DECISION: Instant apply windows should not have a dismissal button. They
> may optionally have a Help button in the lower left corner of the
> dialog.
> 
> RATIONALE:
> 
> The primary incentive for not having buttons on instant apply windows is
> it makes the distinction between instant apply and non-instant apply
> settings pages clear. It also is more accomodating of the general I-A
> implied by instant apply property pages, and emphasizes their imodality.
> This re-enforces a "good" conceptual model for users as it emphasizes
> the direct manipulation aspect of instant apply windows. Rather than
> asking the computer to make changes (i.e. a "dialog" with the computer)
> users are conceptually put in direct control over properties.
> 
> There have been two primary objections to this approach:
>   1) Users will be confused and not be sure how to dismiss the dialog
>   2) Some window managers / themes will not have close buttons

The main objection I find is that:

    3) The window manager close button, when available, is normally a
       much smaller and more difficult to hit target than an
       application-supplied close button would be.

... but there's also the "accessibility objection":

    4) The WM close button is a more difficult control to access using
       standard keyboard navigation techniques.

> Actual objective:
>   Determine if instant apply "dialogs" without button cause problems for
> users who are not experienced with them. The MacOS control panels were
> chosen as the test material because they represent a fairly easy to use
> instance of instant apply preferences.
> 
> [ ... snip ... ]
>
> Conclusion:
> 
> 5 out of 7 users had no problems operating a buttonless instant apply
> dialog. The other 2 users only experienced diffulty because they did not
> recognize MacOS' close button. Although it took them longer both of
> these users eventually guessed the right button.
> 
> My conclusion from this is that its not a learning problem, that people
> pick up on it very quickly, and most people don't even notice.

This objective and conclusion sound as though you've already decided
what outcome you'd prefer.  You conclude that omitting the close
button is the way to go because it only caused difficulty for two of
the test subjects, and they were able to overcome the difficulty
caused by this style of interface.

A more balanced test might have been to set users in front of an
instant-apply dialog with both window-manager and application-suplied
close buttons, and record how many used the large, well labeled
application-supplied button vs how many used the tiny, cryptically
identified window-manager supplied close button.

I suspect this test would have yielded much different results.

-- John Kodis.