The Chinese philosopher Confucius (551 BC to 479 BC) is well known for
his wise and enigmatic sayings. Less well-known are the sayings of Confuseus which are less wise but perhaps more enigmatic.
In fact so much less well known is Confuseus that we
neither know when he said what he said, nor whether he was Chinese, Japanese or
some other ese. However,
what is known is that many of his sayings have pertinence to statistics and not
a few of them have impertinence also. Some examples are reproduced below.
Statisticians are always tossing coins but do not own many.
When you have finished your chi-square it is polite to ask to leave the
table.
There is only one way to truth but ANOVA often has more ways.
An idiot with a computer is often more powerful than a statistician with
a pencil.
The study of many subjects brings enlightenment but statistics brings
significance.
Biostatisticians are always talking about power but do not have any.
Mathematics has lemmas but statistics has dilemmas whereas
Parallel trials meet at infinity but this is more patients than you can
study.
Physicians like within-patient studies but statisticians get cross over
them.
Beware! The triangular test is not used for hypotenuse testing.
Statisticians expect the average but on average people do not expect
statisticians.
Do not give up hope; the statistician may be wrong.
Sit long enough by a sequential trial and eventually significance will
come floating by, unless you take steps to sink it.
The journey of a thousand miles begins with a single step but you will
not get far with stepwise regression.
Do not worry about uniformly most powerful tests; they do not worry
about you.
If you look for a trapezoidal rule in a circus you will be disappointed.
Multivariate analysis will bring you many dimensions but you will not
enjoy them.
Beware of epidemiologists for they are confounding.
You may find pharmacokinetics hard to say but you will find it harder to
do.
A statistical package is not as interesting as the other sort.
Statisticians worry about interactions and this often makes them lonely.
With four parameters you can model an elephant and with five you can
make it wave its trunk but Walt Disney can make it fly.
Even if you have a significant relationship with a statistician you may
not find it relevant.
Chief executive officers are not as clever as they think they are but
they are a lot richer than statisticians.
Statisticians seem to talk double Dutch but parameter estimation is all
Greek to them.
A medical writer is more useful than a book doctor but not much.
When considering types of squares do not count beyond
two.
Many people use statistics as a dog uses a lamppost; beware of p-values.
Bayesians are always shrinking things but they are growing.
A statistician can tell you how large your trial will be but you will
not like the answer.
Enjoy the result you have found by exploratory data analysis, for you
will not find it again.
Statisticians are more rigorous in interpreting statistics but physicians
are more imaginative.
If you get involved in measuring quality of life you will find that
yours deteriorates.
Biostatisticians talk much of censored observations to physicians but do
not censor their observations when talking about physicians.
Clinical data managers are very useful; they make statisticians seem
exciting.
Frequentists think that
it is the thought that counts whereas Bayesians count the thoughts.
You can't beta good power calculation.
If statisticians could communicate it would be in German.
"Degrees of freedom", sounds exciting; do not be misled.
You will find BUGS very simulating.
Beware: an alpha spending function is not a "class A"
bean-feast.
If you ask a statistician many questions expect smaller answers.
Bayesians think that they are, "wacky, zany and generally
interesting," but so do accountants.
Frequentists believe
that the long run is important but in the long run we are all dead so you need
pay no attention to them.
The boundaries approach, although neither cricket nor golf is a load of
balls.
Further
Sayings of Confuseus
Little is known
about the enigmatic Confuseus except that this sage
is not of our thyme. However, what is
known is that many of his sayings have pertinence to statistics and drug
development and not a few of them have impertinence also. Unfortunately, some
more of his sayings have now been discovered
Some people will not say boo to a goose and others
will not say shoe to a bootstrap.
Do not expect to find a linear model on the catwalk.
Bioinformatics - so many syllables, so many data, so
few results.
Relational databases have less to do with dating
agencies than you might think, although those who know most about the former
are most in need of the latter.
Nobody could get the developers of BUGS under the
trade descriptions act.
Micro arrays should be welcomed with micro-hoorays.
"In silico
development" - it sounds good doesn't it?
Acquisitions department: all about finding potential
businesses, buying them and making money out of them. : "Two out of three ain't bad" (Meatloaf)
An elephant never forgets and neither does SPlus, unless you wash its brains regularly.
Neural networks: Developed by analogy with the
workings of the brain to solve brainy problems. Surprising, isn't it?
Theranostics. The fusion of
therapeutics and diagnostics to create an exciting new word.
Theranostics. A subject with all the relevance of
diapeutics
It is not true that marketing departments are
unscientific. They work in sell biology.
Multivariate analysis: such pretty pictures.
You can try make a model
statistician if you want but there will always be residual deviance.
There is something fishy about Poisson regression.
Remember to use the dscale option.
If you think that working with SAS involves flying regularly
to
Between you and I, that's very bad grammar and the two
of we should avoid it.
One who does not learn to program will suffer from
mouse-click.
Stepwise regression: regression certainly, and many
steps but wise?
What is a query?
Jacknife of all trades master of none.
Bayesians: they know that you're wrong but can't tell
you what's right.
Mathematicians only deal in exact truth and so cnnt mk infrncs.
The average person is a miser because they are very
mean.
When spring arrives gardeners get out their weapons of
moss destruction.
If you think drug development is important it is
probably just a phase you are going through.
If one can launch a ship with just a bottle of
champagne how do you explain the hospitality budget of the marketing
department?
You chose to use procrustes
analysis? It is not stretching a point to say that he who make
his bed must lie in it.
If they are opinion leaders, why are they told what to
say?
How to win friends and influence people? You took the
wrong step by becoming a statistician.
If marketing did not exist, liars would have no
purpose in life.
It would be logical if the first author wrote the
first draft.
If you think a spreadsheet will help you manure your
crops please wash yours ears with soap. It does not excel at this.
How can deviation be standard?
To resort to regression is to take a step backwards.
It is amazing how some people can confuse association
with causation. They should apply to join the ASC (American Statistical
Causation).
If you were exchangeable about spelling, you would
also be changeexable.
Little is known
about the enigmatic Confuseus except that if you
study him you will not get wise but 'whys'.
However, what is known is that many of his sayings have pertinence to
statistics and drug development and not a few of them have impertinence also.
Unfortunately, yet more of his sayings have now been discovered.
If you wish
to compliment your boss on running a tight ship, first check that you do not
have a spendency to toonerise.
If you
apply regression techniques to classifying throat diseases are you putting the
CART before the hoarse?
The jury is
out on self-designing trials.
Why is a
shredder like dichotomisation? When it comes to destroying information, it only
depends which doc you meant.
Clinical
trials are supposed to reduce doubt but unlike legal ones they rarely lead to
convictions.
Mssng dt cn b hrd t ntrprt but nt lwys mpssbl.
If the
statistics of Bayesians is all about belief, why are they so sure of
everything?
If you were
jumping out of an aeroplane, a golden parachute would be as much use as a golden
handshake. Perhaps you should buy the CEO a skydiving trip.
If you knew
the answer, why did you employ a statistician?
So you
think double data entry is a useful contribution to quality? Is this why you
prefer Cochrane Collaboration meta-analyses to those from the pharmaceutical
industry?
However
much you hate the marketing department they love themselves more.
Did you
know that 'Yes, main bias' is an anagram of Bayesianism?
Did you
know that 'Queer misfit' is an anagram of frequentism?
No? That's because it isn't.
Don't you
hate the way those meta-analysts go on and on?
Why is the
boardroom like a bullfighting arena? You don't know? Ask the man who cleans the
latter.
Nonparametric
methods stink or at the very least are rank.
So you want
to teach statistics to physicians. Would you like to find out that a child was
flying your aeroplane?
Lift door
closing. Physician will stop with hands, surgeon with head,
statistician will say 'we mustn't confuse trends with evidence'.
How do you
get nightwear from a mathematical constant and a work of fiction? Easy! Just
say PI-JAMA.
Did you
know that a stylos was a column with a saint at the
top? (See http://www.newadvent.org/cathen/14317b.htm. ) Just like an academic editorial, eh?
Mirror, mirror on the wall
Who are whitest of them all?
For I've a paper I've to write
It must be checked by one who's 'white'
My sins are black, I can't be trusted
Without whitewash my paper's busted
To academics, pure as snow
He who'd publish now must go.
If they say, 'yes,' it's in the bag
Unfortunately there's one snag
No academic works for free
And so you'll have to pay a fee
Once paid, they're in the dealer's pay
And none will trust what they then say
So to sum
it all up, what is meta-analysis?
How many
real windows do you know that suddenly freeze and won't move or shut themselves when you don't want or expect it?
It's all a bit of a pane.
The game about management in the pharmaceutical industry? Snakes and
ladders. The game about statistics in the
pharmaceutical industry? Snakes and adders. The
game about marketing in the pharmaceutical industry? Vipers.
Drug development pipeline: a metaphor taken from the petroleum
industry.
Strange! If you had a pipeline carrying oil you would be very alarmed if only
10% of what you put in at the beginning got out at the end.
Drug
development sieve, perhaps?
Analysis of variance - sum of the squares. Statisticians - some squares.
Criminal
pathology - medicine based evidence.