[thelist] Demographics and market research (was The ClientDoesn't Know His Audience)

[Luther, Ron]

john at johnallsopp. asked:

>>So you're the guy to provide us with lots of sources of research on these
>>matters. <cheekyQuestion> Do you have any links, authors, researchers,
>>departments or anything you'd like to share?</cheekyQuestion>

Hi John,

It's been a very long time since I've worked in that field.  It was 
fun, but 'primary research' (the folks who actually design and conduct 
their own research studies) is a pretty narrow field - not a lot of 
jobs there nowadays.

The best practical 'how to' guides I've seen were the three ring binders 
that went along with the courses developed by AT&T's Analytical Support 
Center in association with Bell Labs. Those probably aren't in the public 
domain, but given the historical impact of the telephone industry on 
applied statistical research there probably are a heck of a lot of white 
papers available and/or textbooks from old phone company folks.

For textbooks, our group relied heavily on Cochran (I think it was Robert) 
for the theory, Ed Deming (mostly for our sick and twisted idea of fun), 
and Taro Yamane (which contained a awful lot of errors - but was the most 
accessible of the lot). 

For authors I'd suggest reading anything by Gene Woolsey out of the 
Colorado School of Mines. He's very practical, irreverent as all get-out, 
and generally a load of fun.  (He's a heck of a speaker if you get a chance 
to see him in person as well.)

There are/were also a lot of professional societies and publications: In 
the US these would include the American Statistical Association (Journal 
of the ASA), ORSA/TIMS - Operations Research Society of America and/or 
The Institute for Management Science (Interfaces used to be a very readable 
publication - I would read it for the Gene Woolsey articles - it should 
be available in major libraries), and the ASQC - American Society for 
Quality Control.  Attending some of their luncheons/functions should put 
you in touch with folks 'working in the field' in your local area.

Lastly, I would also suggest 'interviewing' research vendors in your area. 
I used to keep a file on my impressions of the vendors I would hire for 
fieldwork. If I had a meeting in a different city I might fly in half 
a day early so I could interview/inspect research facilities for vendors 
in that area ... just in case I might need some work done there in the 
future. You should be able to find these folks in a 'business-to-business' 
Yellow Pages directory.

Was that the kind of thing you were looking for?

There is *plenty* of literature of the 'business mistakes made because they 
didn't understand their market' variety.  Often light and very entertaining 
reading.  (Any researcher you run into will also have their personal 
'war stories' - some of those can be a lot of fun too!) {Like the Ford 
researcher I ran into who gave me a long and captivating story explaining 
in detail just how and why the Ford Maverick was 'deliberately' designed 
to be 'so ugly no one would want to buy one'.}  ;-)

... or were you looking for more info on what makes a project 'weird'? I 
was once asked to estimate and separate the curtailment impacts of a price 
increase and a change in pricing structure. 

<hypothetical example of 'confounding' price increase and structural change>
Suppose instead of walking onto the lot and paying $50,000 for a new Vette 
this week, next week you'll be asked to pay $26,000 for the motor and 
$26,000 for the rest of the car, oh and by the way, no - we won't sell them 
separately. 

Now estimate how many people who would have bought them last week now walk 
away. Now determine how much of that is due to the price increase and how 
much is due to the structural change in pricing. Fun, eh?
</hypothetical>

I swear I though some buddies of mine in Philadelphia (who used to work 
for the US Census - so they were pretty good stat folks) had done something 
similar in the past so I proposed using mall intercept fieldwork within a 
two-way factorial experimental design to provide new coefficients for an old 
econometric model we had lying around the place. Then I called my buddies 
who said "What the hell are you talking about Ron?".  Ooops! Fortunately it 
turned out to work well anyway.


RonL.

<tip type="empty recordsets made friendly" author="RonL.">
Got a form/report that gives users all kinds of controls to monkey 
about with before executing?  Sometimes it comes back with no 
records?

* Sure, it's easy to just build them a results page with a single 
row of THs showing the column headings for the data they didn't get. 
But is that 'friendly'?

* It's not that much more effort to toss in an "if bof=eof" or "if 
rowcount() = 0" and direct them to a "Sorry, the data contains no 
records for the selections you made" page, is it?

* It's not much more effort to have that 'error' page list for them 
the selections they made that resulted in receiving no records.

* It would be more effort, (because you'd have to build a little 
side table and some logic), but in some cases you could actually 
display some explanatory text explaining _why_ they didn't get 
any records for their section: "Sorry, there are no shipment 
records available because we don't actually sell WOMD to _that_ 
country."

Just some thoughts.

</tip>