Date: Mon, 29 May 1995 22:42:58 -0400 (EDT)
In-Reply-To:
Message-ID:
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Status: O
X-Status:
On Mon, 22 May 1995, S. Finer wrote:
> The problem with using web forms for surveys is not the forms per se.
> The problem is that you limit your respondents to just people who use the
> WEB, a growing but minor subset of general users.
It depends on what type of form you use and what you're trying to survey.
I've had a few survey experiences that were so unpleasant (300 3D-icon
pushbuttons per page!!) I didn't even try to complete the second page.
> An even more critical problem is the volunteer nature of those going to
> the survey URL. Unbiased samples are a critical issue in valid survey
> research, and they are an issue that is habitually disregarded in many of
> the online surveys.
One of the great jokes is the way statistics are used in the entire
computer industry. Novell has 70% of the network-operating-system
market. Unix servers don't count, TCP/IP packages downloaded from the
internet don't count, number of users actually using IP vs IPX doesn't
count. According to NetWorld, this is an unbiased survey. :-).
> The interpretor has NO IDEA what population is
> actually being surveyed, or what relation it has to the superset
> accessing population.
Short of rounding up the general population and forcing each one to
answer your questions you have NO IDEA what population is being surveyed.
A phone driven outcall survey will get different results at 7 PM on
tuesday than 8 PM on friday. The "random population" of people who have
nothing better to do on Friday night than sit around and talk to some
pollster is likely to be looking for the "Magic soap" that will make them
irresistable to members of the opposite sex. Hence the Nielsen ratings.
Can you imagine trying to plan a political campaign or an economic
forcast around that sample?
>For COMMERCIAL purposes, as opposed to feature
> article purposes where a financial investment is not at risk, the
> volunteer bias is unacceptable.
Actually, I would be much more interested in finding out that there were
7000 people who wanted to pay $100/month for up-to-the-second economic
information than to find out that 99% of the "randomly sampled"
respondants wanted their news for free. This would at least create the
basis for a million dollar/year budget. Even if only 20% of those
respondents came through, I could make a tidy profit.
The internet isn't a single generic "population" which will soak up any
type of information you want to shoot at it. The internet is about
20,000 "cities" organized by specific interests and commitments.
Individuals commute freely between the towns. Some of the neighorhoods
are a bit seedy (alt.hate.* alt.white-supremacy, alt.sex.*...) but you
can't be forced to go there either (a cross-post might catch your
interest, but it comes into your field-of-view based on your interests.
> Well whadda ya know......It looks as though the ala carte pricing as
> independent entities, as Bill Densmore, myself, and others described,
> just might get a shot after all, eh? 8-)
Yes, I wouldn't mind paying "penny-a-page" rates for something I could
buy in a "Weekly Magazine" (up to 500 pages/week for 5.00). I might even
pay $20/week for some magazines (PR-Newswire, Software, Stock
Quotes/Reports)....
> On Mon, 22 May 1995, Eric K.
> Meyer wrote:
> > I've just converted several of the bread-and-butter surveys and
> > other features offered on NewsLink into HTML forms.
>
> > At one time I thought that