Re.: Use of online color

[Geoff Hart <geoff-h -at- MTL -dot- FERIC -dot- CA>]

     Dan Wensel wrote <paraphrase> to ask about the use of color to
     distinguish mandatory fields and fields that have been edited on
     certain windows.

     There are many problems with using color, but to be brief, you can
     still use it if you're careful.

     Problem 1: Colorblind users: If you use color, verify that the same
     display works well on a monochrome screen. (This means that the
     display will work even for most colorblind users.)

     Problem 2: I find that color helps me label things primarily in
     drawing/painting programs and games. In "fill in the blanks" forms,
     it's rarely useful and generally doesn't meet my aesthetic standards
     Not that I'm Mr. aesthetics, but people won't always like your color
     choices. If you use color, let users change the defaults.

     Possible solution: Rearrange your screen display so that mandatory
     fields occur together, with the field names in bold type. Put optional
     fields elsewhere, either under a dialog box (Windows convention) or in
     a separate grouping (Information Design convention). To show fields
     that haven't yet been edited, consider using a light grey version of
     the type (say, about 2/3 of full intensity), and change this to full
     black (100% intensity) once the text has been edited. Finally, for
     fields that _must_ be edited or filled out, consider adding a tiny
     checkbox beside the field (always in the same place, say, to the
     left): once the field has been filled in, add a tick or X to it
     (Windows convention); if the user tries to leave the data entry screen
     before all boxes are ticked off, pop up a friendly message that states
     "all fields with check boxes must be filled in before finishing this
     screen" (be friendlier... I was in a hurry!). Note: Combining these
     techniques will be even more effective.

     Final note: You don't need a consultant. Try a few variations, then
     ask some of your users to test it for you. Usability testing can be
     cheap: if your software is for accounting personnel, ask your
     accountant to try it out; if it's for clerical staff, ask your clerics
     (!) to try it out. They'll find out the problems real quick. Once
     you've eliminated obvious problems, try a group of real users and see
     what they think. TO be statistically valid, you should aim for 30 or
     so people (for t-tests etc.) typical of your audience; for better
     validity, stratify your test among the main groups of users (e.g.,
     novices, experts) and aim for more than 30 in each group (for z-tests
     etc.)

     --Geoff Hart  #8^{)}   <--- effective information design