Subject: Re: Stats From: "Eric Meyer" Date: Wed, 18 Sep 1996 01:04:54 -0500
How the Web Was Won
Subject: Re: Stats From: "Eric Meyer" Date: Wed, 18 Sep 1996 01:04:54 -0500

On 17 Sep 96 at 23:34, Christopher Harper wrote:

> Eric, I am trying to find a baseline.  That's the problem I have.  What
> data exist about scrolling other than conventional wisdom?

To use some research-speak, this is an operationalization problem.
If you operationalize scrolling as exploration of the offerings vs.
consumption of the offerings, as Eye Trak experiements have done in
print, there's quite a bit of baseline data, starting with Garcia and
Stark's Eyes on the News and continuing through the works of the
Institut fur Kommunickations-Forschung Beate von Keitz at
Saarbrucken, Germany. We're talking about what the Institut refers to
as the orientation and, to a greater extent, processing phases, which
occur BEFORE the reading phase is encountered. Garcia and Stark
didn't begin to measure the reading phase and barely got into the 
processing stage, which is where predictable efficiency and various 
Gestalt perceptual theories take over. 

> That may be true.  But we do know that people do not want to jump from
> A1 to B13. 

Yes, but this is a terribly misunderstood point. The decision to read 
or not read is made before a jump is encountered. The dramatic 
falloff occurs not at the jump point but just before the lead. Until Eye Trak, 
we thought a lot more people read leads than actually did. Now we 
know that damned few do. The challenge is not to worry about the, 
say, net 1.5% who stop reading at the jump but rather to worry about the, 
say, 95.0% who never even started reading. Yet we continue to flail 
about regarding such things as jumps because we really can't bring 
ourselves to admit that it might not be the packaging that's a problem; 
it might be the goods themselves. 

> There is some limited research that shows that people -- when searching
> -- only want to see the top screen.   

Research tends to like the parsimonious solution. Perhaps the 
simplest answer is that, if they don't see anything interesting in 
the first screenful, why would they want to trouble themselves to 
scroll further? 

There's all sorts of evidence that, if people really want something
and think it's important, they'll jump through hoops to get it.
Witness the number who actually DO follow through to jumps. Approach
it qualitatively sometime. Watch people read. Even the youngest
reader, if the story is compelling, will have no problem following a
story all the way through. It's very likely that it's the story
that's marginal to begin with that loses its readers on the jump
point. I'm not saying we should concede this loss, but I think
there's something inherently dangerous about constantly assuming that
what we write has to be right, so it must be the presentation, the
circulation department, the newsprint, whatever, that keeps people
from reading it. Eventually we have to
consider whether what we write is really that interesting --
particularly in a society that is increasingly detached from its
institutions, which much of our coverage focuses upon.

> The 10 per cent is a scroll figure, which would be the equivalent of
> reading beyond the first or second graf.  

No. It probably is not a "reading" number at all. People could
scroll simply to see whether there's something else beneath what's
currently on the screen. They might be scrolling to look for an ad.
We don't know that they've read anything. In fact, if print
reading patterns hold true, movement like this is more likely to be
associated with orientation and processing than actual reading. Or,
at least, so says Eye Trak.

> It is indeed foolish to surmise what readers and viewers do without hard
> proof.  With all due respect, kind sir, you are making some leaps of
> analysis here without hard proof.  You and I both would like to have the
> stats necessary to figure out what's going on.  Simply put,  we don't
> because it ain't easy.  

Very true. In fact, there is no known way to measure reading as a 
behavior. We can surmise when reading takes place by watching eye 
movements go back and forth across lines, progressing from one line 
to the next in sequence. But long before that happens in print, the 
eye very often moves it vertical leaps (without the horizontal 
patterns) across the text. The only way an eye could do that online 
is with scrolling. That's why I say there's a pretty good chance that 
scrolling and reading are not the same thing.

Moreover, suppose we had data saying that people who really wanted
to read a story (how we'll *ever* know that for a fact is beyond me)
were dissuaded from doing so because they would have to scroll. What
would we do then? Write shorter and put little chunks on each
screen? Then we'd hear that people didn't want to click hyperlinks to
continue. Should we have one button that, when we decide a story is
interesting, would automatically generate a printout of the full
text? Maybe. But Eye Trak has shown that additional information
gleaned during the orientation and processing phases before reading
may be necessary before a decision to read is reached. In other
words, people will look at the pull quote, graphic, photo, and other
things scattered about in the text, and use that information to
decide whether to read. A system we create of generating a printout
on demand might short-circuit that process by not exposing the
readers to these while they are still pondering what to read. While
the few who chose to read might be happier, the number who started out
reading would be so diminished as to make the gains meaningless.

There are some real problems with online design. I suspect we'll
find that type legibility, more than readability, is a prime
concern, mitigating for use of, say, Helvetica instead of Times.
We'll also almost assuredly find that most online type is set too
wide for optimal reading and that the angle of reading is ill-suited
to broad comprehension. We also probably will find that type should
be a muted amber on a dark background, not black on white. We'll
also probably find that there's less informational choices presented
on any given online screen than within the visual range of any portion of a
standard newspaper page -- fewer hooks and entry points, if you will, 
meaning fewer topics, thereby fewer chances of developing a habit of believing
that it is worthwhile to mull through for interesting "finds."

Yes, we don't have data. But based on established data and theories
of information design, it's my opinion that scrolling is nowhere
near the top of the list of things we should be worrying about. It 
is, in short, a red herring issue that belies blind faith in the 
value of all else we do.

> You and I and many people on this list know that
> we have to come up with these figures pretty darn quick to justify this
> medium.  Actually, you and I don't because we now have the comfort of
> the ivory tower of academic life.  Cheers. 

Actually, online is a bigger income source for me than is the
academy, so I fall into the former group. What worries me is that
there's a tremendous need for a thoughtful research agenda to be
established, but no one is establishing one and no one is showing any
inclination of following through in a methodical pattern. Everyone is
trying to do the same thing -- albeit often half-assed -- and most of
it is no more than stupid surveys that ask people what they think or
feel about something. You're right. We need some absolute data. And
the only way to get it reliably, with any degree of external
validity, is with controlled experiments conducted within real
publications -- something virutally no publication will allow.
Sitting freshman journalism students down in front of a computer,
instructing them to read something, then asking them whether they
liked to scroll or not won't tell me a whit about this. A carefully
structured experiment in which, say, Newspaper X's page is
redesigned so that every fourth reader gets a no-scroll verion, with
every other reader surveyed on comprehension, and all mouse clicks
tracked, would begin to do it. But by and large, no newpspaer will
hear of this, and no one will fund what would be a pretty costly 
experiment. I managed recently to get a little bit of funding for 
some real-world experiments I've been doing, but it's nowhere near 
the amount needed for something like this.

A few of us who've spent large parts of our careers studying
information design could easily put together a damned good research
agenda. In fact, a few of us have worked about half of it out in idle
conversations. But ever some well-positioned people can't get this
stuff done because everyone turns a deaf ear to anything but yet
more insipid surveys. In measuring attitudes, such research is great.
But in measuring real world behavior, it's as valuable as
interpretting cricket chirps. 
+---------------------------------------------------------------------------+
This message was posted to ONLINE-NEWS. http://www.planetarynews.com/o-n.html

------------------------------

End of online-news-digest V1 #790
*********************************


From owner-online-news@marketplace.com Wed Sep 18 14:23:30 1996
Received: from marketplace.com (majordom@marketplace.com [206.168.5.232]) by cnj.digex.net (8.6.12/8.6.12) with ESMTP id OAA03074 ; for ; Wed, 18 Sep 1996 14:23:29 -0400
Received: (from majordom@localhost) by marketplace.com (8.6.12/8.6.12) id HAA02974 for online-news-outgoing; Wed, 18 Sep 1996 07:56:37 -0600
Received: from server.indra.com (server.indra.com [204.144.142.2]) by marketplace.com (8.6.12/8.6.12) with ESMTP id HAA02969 for ; Wed, 18 Sep 1996 07:56:33 -0600