Re: Statistics in Psychology?

In article <lv2dnSXdt9zqogrZnZ2dnUVZ_v6dnZ2d@xxxxxxxxxxx>,
Jerry Dallal <gdallal@xxxxxxxxxxxxxxxxxxxxxxxx> wrote:
Brett Magill wrote:
On Mon, 19 Jun 2006 01:42:48 -0700, Reef Fish wrote:

.......................

There are more data--important data--than fully trained statisticians to
analyze them.

The problem in a nutshell is that the statistics profession as a whole
has not done a good job of coming up with a way to train people
analogous to the training of a PA or NP. Some masters biostat programs
do a good job, but many in the profession take the position that nothing
less than a complete statistical education can produce someone capable
of serious analysis. Maybe they are right. While there are
limits--there's little one can do in a single semester to turn someone
into a serious analyst--I think there are things that could be done to
train a statistical NP with perhaps four semester courses of hard work
spread out over two years. The real trick--as many of the posts in
sci.stat.* demonstrate--is teaching people to recognize when they are
out of their depths, the equivalent of an NP knowing when to refer to an MD.

My answer is that the most important part is missed; we
cannot TRAIN people to carry out the analyses, especially
the ones which should be done and are not being done.

The entire miseducational systems trains instead of
educating. For example, we emphasize how to solve
problems in algebra, and also in calculus, rather than how
to formulate the real problems, which may or may not be
solvable with the devices available in the elementary
courses. As I have stated many times, it does no good to
know how to add if one does not know when, and if one
knows when and how to get the computer to carry out the
how, fine.

To use statistics properly, the user needs to FORMULATE
the probability model. This requires a more basic
understanding of probability than merely the limit of
relative frequency, but something inherent. So at
whatever level, probability is needed. The probability
model, and whatever assumptions need to be made, are not
a matter of statistics, but of the field, and here we have
a lot of garbage going on, such as transformations to
normality (a permanent non-no), and even normalizations
to get correlations instead of covariances; they mess up
the conclusions which can be drawn. If statistics is going
to be used for decision making for action on individuals,
I see no way of avoiding a Bayesian approach, although
robustness (definable by the user, but unlikely to be
evaluated except by a mathematical statistician) can be
used to reduce the prior assumptions.


I am often requested to repost my five commandments. These are
posted here without exegesis.

For the client:

1. Thou shalt know that thou must make assumptions.

2. Thou shalt not believe thy assumptions.

For the consultant:

3. Thou shalt not make thy client's assumptions for him.

4. Thou shalt inform thy client of the consequences
of his assumptions.

For the person who is both (e. g., a biostatistician or psychometrician):

5. Thou shalt keep thy roles distinct, lest thou violate
some of the other commandments.



The consultant is obligated to point out how their assumptions affect
their views of their domain; this is in the 4-th commandment. But the
consultant should be very careful in the assumption-making process not
to intrude beyond possibly pointing out that certain assumptions make
large differences, while others do not. A good example here is regression
analysis, where often normality has little effect, but the linearity of
the model is of great importance. Thus, it is very important for the
client to have to justify transformations.

There are, unfortunately, many fields in which much of the activity
consists of using statistical procedures without regard for any assumptions.
--
This address is for information only. I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Department of Statistics, Purdue University
hrubin@xxxxxxxxxxxxxxx Phone: (765)494-6054 FAX: (765)494-0558
.

Relevant Pages

  • Re: Statistics in Psychology?
    ... While there are limits--there's little one can do in a single semester to turn someone into a serious analyst--I think there are things that could be done to train a statistical NP with perhaps four semester courses of hard work spread out over two years. ... To use statistics properly, ... the probability model. ...
    (sci.stat.edu)
  • Re: YA Airline Memobitch...
    ... Since 1970 <300 people have died due to train accidents in the United ... I pulled those statistics from the FRA and the NTSB. ... ....and those numbers are per billion passenger miles. ... that airplanes have a better track record, ...
    (rec.sport.football.college)
  • Re: 1816 London Victoria to Eastbourne forgot to stop at Three Bridges
    ... If a train doesn't stop at all the stops it is supposed to stop at it ... before it contributes towards the cancellation statistics? ... be a particular train that always runs late and always inconveniences ...
    (uk.railway)
  • Re: YA Airline Memobitch...
    ... safety than trains. ... Since 1970 <300 people have died due to train accidents in the United ... You pulled those statistics out of your ass. ... both forms of transportation are safe ...
    (rec.sport.football.college)
  • Re: Statistics in Psychology?
    ... train a statistical NP with perhaps four semester courses of hard work ... Herman and I agree on our "miseducational system", ... mathematical statistics oriented) to be a MISEDUCATIONAL ... SYSTEM for an applied statistician. ...
    (sci.stat.edu)