NONE: Net and Web Use...
Net and Web Use...
S. Finer (xerxes_at_clark.net)
Wed, 24 Jul 1996 03:06:37 -0400 (EDT)
Replies in context, below.
On Tue, 23 Jul 1996, John Carlton-Foss wrote:
> I attended the MIT symposium, and thought I would put in my two cents worth
> here. My comments are based solely in response to Donna Hoffman's note, and
> do not reflect a direct reading of xerxes' full-context note.
>
> It was very jarring to be at MIT and to learn that a rift had occurred
> between Hoffman and Nielsen. The senior vice president of Nielsen sat right
> behind me in the audience and it was clear that Nielsen was jarred and
> embarrassed by the rift when he spoke. What was not clear at the time was
> who was right, but Nielsen commanded enough respect in that group and their
> study was otherwise so massive and innovative (whether one agrees or
> disagrees with the validity of the assumptions and techniques used) that it
> was not resolved whether their inflated number might be correct.
Not offically, but there was an informal concensus. The problem was the
definitions used to ascribe "use"....which several Neilsen people spoke to
directly in their presentations. If you lived in a household where your
spouse used the net, and you had sent an email, or browsed the web once or
twice in the past, you would have been counted positively. This is a
problem for me, as it tends to inflate the total. Access to the net and
use are different.
> At issue
> was that it found significantly higher numbers than those of any other study
> presented at the symposium.
And also at issue, the fact that a computer industry group had funded the
work unlike other studies.
> My own resolution, and I think that of more than a few others, was that if
> you look at the various results of various studies, and factor in the
> biases, limitations and delimitations of those studies, there were actually
> about 18M internet users.
I do not recall this conclusion, and find it unlikely. It is quite
impossible to say that in total the two groups you mention, home users and
work users, are mutually exclusive. There is a strong overlap, not only
people who use the net in both locations, but people who work at home
also. Beware assumptions that lead to double counting.
> Roughly 9M were individuals and about 9M were
> corporate personnel. Note that these are very approximate numbers. Plus or
> minus a million would be almost expected for error brackets.
and what do you see as the overlap between these two groups, or do you
imagine they are mutually exclusive? There are millions of people who
access both from work and from home, frequently with different accounts.
For advertising, total unique users, people who use the Net at least
semi-regularly, are of more interest. People who access ad sites, or who
purchase online, are of most interest.
I have heard some say that there is at least one user for every registered
host.....this MIGHT be true, BUT, there is absolutely no way that there
is a UNIQUE user for each and every account. This fact inflates total
user estimates unless specific precautions are taken.
My own estimate is that there are between 11 and 15 million regular users
in the US at this time, and approximately 60% access both from home and
work. This suggests that there are many dormant, or near dormant accounts.
[snip]
> closer to the consensus. It is important that there be good reasons for
> major deviations between the various studies. Nielsen's results are
> valuable, even if overestimated, partly because of the assumptions and
> techniques used in their work, and partly because the results make us think
> a little more clearly about what is going on in this revolutionary phenomenon.
>
> > >The CommerceNet study, funded by an industry trade group, as Ms Hoffman
> > >knows, received rather through review at the MIT survey methodology
> > >symposia last January 29-30 (see cite below).
this is a true statement, and hardly hostile......
> >
>
> >Nielsen did indeed present their results, but these results were never
> >reviewed, thoroughly or otherwise.
> THIS IS A CORRECT STATEMENT BY HOFFMAN.
Oh really,.... the participants at the conference offered no thoughts on
the study and its assumptions? Hardly true. All the studies at the
conference were reviewed by the audience attending. And, some were found
more acceptable than others. In fact, the conference moderators, mounted
the stage after each set of presentations, and summarized their findings
and methods, usually with incisive comments for each. Perhaps my
definition of the term "review" is what you find fault with? Eh?
The entire proceedings were reviews of work in different areas of
Inet measurement......
> >In contrast, we have written three
> >papers about the Net and Web demos, and have posted them on our Web site for
> >review and comment. Additionally, we submitted one of the papers to CACM
> >(Communications of the ACM), and it was peer reviewed and accepted for
> >publication.
Sounds like a lot of work. But if the original screener in the random
digit dial survey were problematical, because of definitional problems,
then the extra revisional effort may be for nought. Who was considered a
"user", and who was not, is the first question that comes to mind.
In the context of advertising, the nature of the audience is critical.
USERS should not be people who seldom logon (less than once a month) the
Internet. Mere access to an Internet account cannot be considered a
unique user. And ownership of numerous accounts by a single user, cannot
be double counted.
Access to an account, IMHO, does not a user make. Others may differ.
Personally, I'd want new data at this juncture, not a rehash of last
year's work, last year's respondents,...etc. The character of incremental
new users does not appear to be static, i.e. the mix of market segments
entering the medium is changing.
> >And although Nielsen has criticized us for preparing these written
> >criticism, they have yet to criticize our FINDINGS on substantive grounds.
And I doubt they will, as there is little commercial point in bothering.
It may have to do with a situation for which Americans have a quaint old
saying... "..do not get into a p*****g match with a sk**k..." After all,
they have a business to run. At this point, I suspect they'll just let it
alone and go on to new work, with fresh data, fresh definitions, and
freshly revised team, and perhaps....new clients too. Is Nielsen still
doing work for CommerceNet, BTW?
I have heard that they (Nielsen) will be VERY careful about how they use
academics in future work.
Oh yes, John Carlton-Foss, says he heard no comment from the sponsors
about the Nielsen study. AND HE SAID IT IN CAPITALS!
Gee John, did you attend the meals, walk through the foyers, or chat
with anyone? Perhaps you kept a very low profile? But then, your name
does not appear in the Conference roster either, not that it proves
anything, but....well....
Don't you remember MIT Prof. Roger Hurwitz reviewing the first morning
session where Harkness and DuBois presented Nielsen's work? There was
considerable consternation over the disparity in totals among the various
studies. And in response various researchers spoke repeatedly to the
issue of definitions employed during sample selection...the same issue I
have tried to bring to this list. It is not a trivial issue in surveying.
Finally, I do not use a sig. because I find them too frequently offensive.
