Antiquated Software

[Fairlight]

Is it just me, or did Laura Brody say:
> 	I would bet cold-hard cash in a non-trivial amount
> that a top-level data-entry gal who has been at it since
> Eisenhower was president would have a drop of productivity
> of around 50% or more if she had to navigate through the
> exact same screen with a mouse on some GUI database.
> 
> 	Does anyone have any kind of ideas that would
> help us all out by proving this scientifically?

Design an entry program in both Access (or whatever GUI) and filePro for
the same exact data.  Lay the screens out as virtually identically as
possible.

Using the exact same machine, hire a succession of temps (or get
volunteers) to spend 8hrs doing data entry for three days each.  Half the
day they enter data from hardcopy into fP, half the day they re-key the
same identical data into the GUI.  The first day I'll talk about later.
The second day, have them enter fP in the morning, GUI in the afternoon.
Reverse the program order on the third day.  

The switching is necessary to prevent skewing by "warm-up" and "burn-out"
phases that just plain -will- happen when doing data entry for protracted
periods.  Many people really hit only 4-6 -really good- hours of data entry
out of 8 in a given day, with the mediocre-to-lousy times usually on either
end.

Make sure the hardcopies are in the same exact order each cycle.

Keep track of how many complete entries they get through in each cycle.
The results should be telling when spread over multiple people.

For optimum accuracy, the same exact computer should be used for both,
right down to the keyboard, monitor, and even chair and lighting.  A
windowless room with only artificial lighting is preferred.  The less
variations at any given point, the more accurate it will be.

I think you'd also need to give one warmup day (the very first day of 3)
for familiarisation with both programs, so there's no speed increase on the
second day simply because they're finally acclimated.  From personal
experience, acclimation takes 2-4 hours depending on the complexity of what
you're keying and how you have to key it.  By that time, you should be
cruising, if you're even worth hiring for data entry.  If you allow for
this with the extra day, the second two days should not show a huge
discrepancy in most cases.

Best test I think you'd be able to get.  Don't ask me how to pay for it.
You wanted an idea, and that's about as scientific as I can make it,
accounting for as many things as I'm remembering offhand about doing that
for a living.

mark->