Message-Id: <199904271602.LAA12598@subcellar.mwci.net>
From: "eric nielsen" <entfolks@mwci.net>
To: <pinhole@exploratorium.edu>
Subject: Fw: Mendel spinning in his grave -LOL
Date: Tue, 27 Apr 1999 10:59:28 -0500
----------
> From: eric nielsen <entfolks@mwci.net>
> To: Steven Eiger <eiger@montana.edu>
> Subject: Mendel spinning in his grave -LOL
> Date: Tuesday, April 27, 1999 10:48 AM
>
> Steve --
>
> With respect, I am attempting to explain the statistical reality here,
> *not* offer another damned fool's opinion. I apologize in advance if my
> explaination seems brash; I have but a limited amount of time to spend on
> matters such as this so I'm not trying to be rude, but rather
succinct....
>
> a) the MEAT: see the diagram I posted at
> http://scienceguy.webjump.com/mendel1.jpg to further illustrate this
> point.... The diagram represents the entire univers of possibilities --
> and, yes the Pr(one of each gender)= 2Pr(both male) = 2Pr(both female) =
> ..5 **BUT** plz note that Pr(both the same gender) = Pr(one of each
> gender), because you add the .25 probability that they are both male +
.25
> =Pr( both female) = total of .5 Pr(both same gender) -- there are two
> distinct ways to be the same gender...
>
> now, back to your scenario -- let's say that the person you know is
sibling
> "A," and your are reasonable convinced that his gender is male. THIS
FACT
> ENTIRELY ELIMINATES **ALL** OF THE RIGHT-HAND COLUMN of the Diagram of
the
> Probabilistic Universe (mendel1.jpg) -- this column which represents the
> universe of probability if "A" had in fact been female. He's not. Thus,
> this contingencyis eliminated because it never happened.
>
> ALL THAT REMAINS IS THE LEFT HAND COLUMN of the diagram!! which
signifies
> that "A" is indeed male.
>
> within that lefthand column, you can see that there is a 50% chance of
his
> sibling being male, and a 50% chance of his sibling being female.
>
> quo ergo demonstratum
>
> b) a detail note -- 23 replicates is inadequate to draw statistical
> generalities, FYI. pick the program of your choice & do 2000 -10000
> randomized reps if you want to get valid data about a statistical
> distribution. Excel will have an =RANDOM function or something similar,
> I'm certain.
>
>
> I have flogged this topic with 100's of undergraduates; it seems
inherently
> counter-intuitive to many people. I **DO** know what I'm talking about on
> this one, however, and if your interested in learning, I'm always
> interested in explaining. Oh, by-the-by, I did not register to get the
> pinhole listserve until just today, but I figured I try it for a while &
> if it is too addictive & time consuming I can always un-subscribe. Take
> home point here is that I did not get to see Paul's "eloquent
restatement"
> & naming of this misconception -- could you send me a fwd of yesterday's
> digest so that I'm up to speed on this conversation -- or at least fwd
> copies of relevant comments.
>
> There is an excellent & exhaustive stats text online at statsoft.com,
FYI
> -- you may wish to review the concepts of independent events & marginal
> probabilities. Also, a chi-squared test of independence is implied by
your
> question; reading up on that -- with the assumption that you read to
> **understand the concept of marginal probabilities*** -- may help
> clarify....
>
> bye for now -- please keep in touch -- Eric
>
> p.s. -- I see your're in Bozeman, right? what a great little town! my
> sis-in-law used to live there & we visited a couple of times.... what is
> your field of study??
>
----------------------------------------------------------------------------
> -------------------------------
> NEW ------>>> ICQ # 25792640 for realtime dialog needs
> Eric Nielsen entfolks@mwci.net 319/245-2561
> http://scienceguy.webjump.com/
> comments appropriate to the public domain may be aired at:
> http://www.InsideTheWeb.com/mbs.cgi/mb165426
>
----------------------------------------------------------------------------
> -------------------------------
>
>
> ----------
> > From: Steven Eiger <eiger@montana.edu>
> > To: eric nielsen <entfolks@mwci.net>
> > Subject: Re: Re: pinhole probability
> > Date: Monday, April 26, 1999 6:03 PM
> >
> > Eric, I still am attracted to my explanation, especially after Paul
> quite
> > eloquently restated it, and even gave it a name. I decided to do the
> > experiment; I have just tossed two coins 23 times and I get roughly
twice
> > as many head-tail combos as either two heads or two tails. If I assume
> > this is a good model for two sib families in the entire species, and
> choose
> > one result, eg. heads to be female, then there is twice the likelihood
of
> > her sib to be male/tail. If there were an infinite number of sibs, you
> > would be right, a situation you probably assume in your work. Here
there
> > are just two, and thus we can toss out the male-male possibilities,
which
> > changes things. We also might do the experiment differently where we
> toss
> > one coin and then another, but that does not seem the proper model as
we
> do
> > not know which sib is younger, and that would have us lose the male
> first,
> > followed by a female possibilites.
> > Steve
> >
> > >being both a biologist & a statistician, I think I'm qualified to take
a
> > >crack at this one.... Steve below does *not* understand (no offense
> > >Steve)
> > >
> > >toss two coins, coin A and coin B. As B lands, you cover it so that
you
> > >cannot see the result. let's say that A comes up heads. What is the
> > >proability that B is also heads, Pr(B=heads) ??? This probability is
> 50%,
> > >or Pr(B=heads) = 0.50. They are independent events, and the result on
> coin
> > >A has no impact on the result on coin B.
> > >
> > >Now let's move on to real-life critters. Mendelian genetics begins
with
> > >the model that genes (and correlated traits) associate randomly during
> the
> > >production of gametes. Under this most basic model, the probability
of
> the
> > >second sib having the same gender as the first is exactly 50%. They
> (the
> > >two sibs) are independent events, just like the coins tossed above.
> > >Period. Full stop. Do not procede further until this makes sense.
> Even
> > >your question is essentially *statistical* then the answer is in the
> > >paragraph which you just completed!!
> > >
> > >the following integrates the stats with a bit of real-world
biology....
> > >
> > >This is merely the beginning model, however -- two genes which reside
on
> > >the same chromosome do *not* associate randomly!! They are linked,
both
> > >physically & probabilistically. I should note that this "genetic
> linkage"
> > >caveat does *not* apply to gender, because gender is determined by the
> > >presence or absence of an entire *chromosome,* so genetic linkage
> cannot
> > >cause a deviation from the Mendelian prediction of a gender ratio in
> > >siblings/offspring. I *do* know of the existance of genes which
> > >influence the gender of offspring in some animals, such that the sex
> ratio
> > >of offspring is skewed away from an expected Mendialn ratio such as
> 50/50.
> > >Assuming that analogous genes exist in the human population (I believe
a
> > >teneable assumption), you could argue that -- for example -- the
> presence
> > >of a one male sib increases the probability that there is a
> "male-favoring"
> > >trait in the family linage, therefore the probability that other sibs
> would
> > >be male as well is greater, but I think that it's rather grasping at
> > >straws. Realistically, I doubut you would detect such "gender biasing
> > >genes" without a sample size in the at least 100's if not 1000's of
> > >families. Might be an excellent thesis in here somewhere -- could
work
> > >from genealogy club notes & historical records to establish a
> longitudinal
> > >study....??
> > >
> > >ok, I've spent quite a bit of time on this; hopefully it clarifies.
any
> > >questions: ask. by the way, gender can be determined by
environmental
> > >vaiables as well as genetic composition -- e.g. cichlid fishes.
> > >
> > >Eric
> > >scienceguy.webjump.com
> > >----------
> > >From: Steven Eiger <eiger@montana.edu>
> > >To: Pinhole Listserv <pinhole@exploratorium.edu>
> > >Subject: Re: pinhole probability
> > >Date: Friday, April 23, 1999 12:02 PM
> > >
> > >>Hey pinholers, a probability question:
> > >>
> > >>Let's pretend that you meet a person and she says that she has only
> > >>one sibling. What is the probability that her sibling is a boy? How
> > >>about a girl? I'm pretty sure that it is not 50:50....
> > >>
> > >>(So, in real life, if you meet a woman with one sibling, is there
> > >>really a better chance of her having a brother than a sister?)
> > >>
> > >>Stumped in Oakland
> > >>
> > >>--
> > >>‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡‡
> > >>
> > >>Ian Bleakney;
> > >
> > >I am not a statistician. But my guess is to treat this like flipping
> > >coins. There are four possibilities two heads, two tails, a head
> followed
> > >by a tail and a tail followed by a head. Since we have eliminated one
> by
> > >knowing the sex of one sib, and we do not know whether she was the
first
> or
> > >second sibling, I would guess that there is twice the likelihood of
her
> > >sibling being of the opposite sex, relative to the same sex. I would
> also
> > >guess that there is a 50% chance I am right about this. I have had
this
> > >explained to me several times and it never seems to stick. Steve
Eiger.
> > >
> > >Steven Eiger, Ph.D.
> > >
> > >Departments of Biology and the WWAMI Medical Education Program
> > >Montana State University - Bozeman
> > >Bozeman, MT 59717-3460
> > >
> > >Voice: (406) 994-5672
> > >E-mail: eiger@montana.edu
> > >FAX: (406) 994-3190
> > >
> > >
> > >
> >
>
>---------------------------------------------------------------------------
>
> > >
> > >To unsubscribe from pinhole, send an email to
requests@exploratorium.edu
> > >with the words 'unsubscribe pinhole' (without the quotes) in the
SUBJECT
> > >of the email.
> > >
> > >To subscribe to the digest and only get 1 combined message a day, send
> an
> > >email to requests@exploratorium.edu with the words 'subscribe digest
> > >pinhole' (without the quotes) in the SUBJECT of the email.
> >
>
>---------------------------------------------------------------------------
>
> > >
> > >
> > >
> > >
> > >----------
> >
> >
> > Steven Eiger, Ph.D.
> >
> > Departments of Biology and the WWAMI Medical Education Program
> > Montana State University - Bozeman
> > Bozeman, MT 59717-3460
> >
> > Voice: (406) 994-5672
> > E-mail: eiger@montana.edu
> > FAX: (406) 994-3190
> >
