----- Original Message -----
From: dennis roberts <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, June 29, 2000 12:32 PM
Subject: dissertations = hack jobs
> it appears that da heiser said something like(if i am in error, forgive
me):
>
>
> > > .............................................................
> > > It is now too late to do a good job. Your disertation is now like all
the
> > > others, a hack job. Do mumbo-jumbo to get the disertation approved.
> > >
> > > DAHeiser
'''''''''''''''''''''''''''''''''''''''''''''''''''''
Yes I said that. I did not intend that all dissertations are hack jobs, but
many are. My frame of reference is a dissertation that is based on
evaluation of collected data, not a mathematical development or an exercise
of an alternate view on someone's else's conclusions. It also is a
dissertation in a non-stat department (chemistry, physics, medical,
psychological, marketing, economics, social studies, etc.) in which stat is
used as a tool.
You have to remember these points.
1. Go down to the basement of your university library and count the many
shelves devoted to dissertations and thesis. How many of them ever show up
as references in papers? How many are important? how many become journal
articles? How many have good data and have a good result? Just guess the
proportion that were written just to get the degree.
2. Look at the approval process. Student writes a dissertation. Approving
professor usually is time pressed and does not devote enough time to
guidance. Student has to understand the professor and the subject, and this
is an evolving situation on a time scale of semesters. Usually the two are
at different poles when the thing is started. After about 80% is done the
professor, on the backstretch, suddenly says to the student, you should do
an ANOVA or a regression or apply the XXX test, which the student has very
limited knowledge of. By the time that graduation approaches, the time
crunch is on. It is a race at this time by the student to get something in
writing that the professor approves of so he/she can graduate. The focus is
on approval, not "scientific" accomplishment. (I think that peer review as
it is currently done "stinks". A peer reviewer should be involved in the
front end, not just at the tail end. Just like with a horse, you avoid being
close to the tail-end. If you are sticking with the tail end, be prepared.)
3. Sit on EDSTAT for several years, and see the constant stream of "help".
"help"........ I have data and want to do some statistical stuff on it. Give
me a good stat software package, or "how do I do regression" or how do I do
an ANOVA, etc. etc. etc. It is always putting the cart before the horse. How
can I not conclude that the result will be one of those forgotten "hack" job
dissertations that just collect dust in the basement.
4. As I said in my message, a good dissertation requires putting the horse
before the cart, Determining an objective, a hypothesis, a conjecture, a
proposed relationship, validating/rejecting a proposed causal relationship,
identifying variables, ranges, significances, probable errors, etc., etc.
Look at it realistically before you have put a lot of time on it and the
time crunch is on. The data and arguments should show or prove the
conclusions without having to rely on whether the p value is 0.044 or 0.052.
(After R.A. Fisher). You have to start with some physical reality, develop
some approach, collect some fragmentary preliminary data, (even some
hypothetical invented data) try out some preliminary stat analysis and see
where the problems and holes are, including your knowledge and
understanding. Don't suddenly call for help on regression analysis at the
last minute. There are just too many ways to do regression.
5. If there is no front end work, the dissertation may be well written, and
have convincing logic, but it has no foundation. It is an "after the fact"
job. It is a desperate attempt of recovery.
DAHeiser
P.S. My response tonight was assisted by some "Black Butte Porter" from the
Deschutes Brewery, Bend Ore. Its really good stuff.
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
