[IxDA Discuss] Testing a hypothesis

[Anderson, Douglas W.]

Hi Robert,

I chimed-in earlier in the thread and have followed it with varying interest as it has evolved. Thanks for your pugnacity, which has helped keep it going.

When I read the bit of your recent message (below), I was struck by an idea. (I wasn't hurt badly, needed only a few stitches to close the wound. ;>)

Is to say, "the further away the user gets from the beginning of a task" the same as to say, "the closer the user gets to the end of a task"? In a finite-length task that would seem to be the case.

If so, (all other variables held constant) wouldn't the *odds* of completing the task improve as the user gets closer to the end? The user would have survived more and more of the "difficulties" that might have caused them to abandon the task. It's analogous to my odds of living to age 70 being better the nearer I am to age 70 - I didn't die from all those things that might have killed me along the way.

Perhaps what you have observed is that the longer a task is, the lower the probability that a user will complete it (where task "length" might be measured in time, "difficulty," frustration, irritation, ugliness, roughness, disengagement, and on any number of other dimensions). 

Within a task, the probability of task completion would have some value at the start of the task and a different value at each step along the task flow. The values would increase as the user progressed through the steps in the task. Some increases would be larger than others because some just-completed steps would have been more "difficult" than others.

Between tasks, a "longer" task would have a lower initial probability of completion. But the measurement of task length would have to capture the dimensions that are (most) meaningful in the system-user context of use. Seems to me that's where the attempt to measure becomes very challenging.

FWIW,
Doug Anderson
Mayo Clinic
Rochester, MN

Opinions expressed are necessarily mine, not necessarily those of the Mayo Foundation.

Original message (fragment):
Date: Sun, 20 Aug 2006 21:29:13 -0700
From: "Robert Hoekman, Jr." <rhoekmanjr at gmail.com>
Subject: Re: [IxDA Discuss] Testing a hypothesis
To: "Jared M. Spool" <jspool at uie.com>
Cc: discuss <discuss at ixda.org>

<snip>
Ah. The objective is to find a way to measure in a more scientific way
what I originally observed, which is that the further away a user gets
from the beginning of a task, the more difficult it was to complete.
I've been trying to restate "further away" as number of steps and
complexity of steps and so on. I've seen this quite a few times, and
we're basically sorting out how this might be turned into something
reliable and provable, if that can in fact be done.
<snip>