Re: pizza

[Mike Granaas]

Interesting point.  Yes, if the Ss do something other than a random guess
the binomial model would be violated.  The question then becomes what
would they do if they are uncertain?  I suspect that they would fall back
on visual inspection...which piece appears to be different than the others
(less green pepper, more browned, etc)  Such information is probably
relevant often enough that "guessing" would be well above 1/3.

Using blindfolded Ss will deal with that problem, and gets us back to
the question that Dennis is asking.  I'm guessing that rather than going
through some sort of a systematic process (e.g. binary decision for the
first piece, progress to second piece only if first piece was judged
"same".) Ss will in fact do something more like guessing.  Only they
will condition their guesses such that if they picked slice A as different
on the previous trial they will first consider slices B and C on the
current trial (they will actually avoid selecting the same slice position
on sequential trials).  Furthermore they will try to equalize the number
of position choices they make across the experiment so that they choose
each of A, B, and C three times and one of those a fourth time.

This leads to: trials are not independent.

The question remains, are the trials not independent in any way that
matters for the purposes of binomial probabilities?  Off the top of my
head, I'm not sure.  I am aware of, through secondary sources, studies
where any amount of guessing about random events (eg coin tosses) results
in a lower number of correct outcomes than simply making the same choice
every timethat is always choosing "tails" will be superior to making
choices between "heads" and "tails" on different trials.  

This suggests that, at the very least, the correct value for p is less
than 1/3.


Michael

On Fri, 23 Feb 2001, dennis roberts wrote:

> a concern i have in this situation ... and why i posed the question is as
> follows
> 
> since it is a taste test ... Ss will taste the pizzas ... so, the notion of
> just selecting ONE and saying it is different seems not a reasonble scenario
> 
> so, what would a resonable guessing scenario be? one might be that ...
> after tasting and retasting ... the S says to himself/herself  ... i just
> cannot make a choice ... i really don't know the difference ... BUT, he/she
> has to make a choice ... those are the rules ...
> 
> so, if that were the case ... let's say that the strategy he/she adopts is
> to flip a mental coin ... if it is heads, call the first pizza SAME ...
> and, if tails ... call it DIFFERENT ...
> 
> now, if the first turns up heads ... then there is another piece to do the
> mental flip for ... so the second piece gets the second flip ... assume it
> too is heads ... and is therefore called SAME ... 
> 
> then, there is NO random choice for the third ... it has to be DIFFERENT
> ... the third slice decision in this case is NOT independent of the second ...
> 
> but, what if the first slice mental flip came up TAILS ... then for it, it
> is called the different one ... but automatically and out of the control of
> the S are the decisions for the other two ... they are both SAMES
> 
> i claim that in this situation ... the decisions for all three are NOT
> independent decisions ... therefore, it does not satisfy one of the
> conditions for the binomial to be a correct model ...
> 
> if the strategy were to simply flip a three sided coin ... with sides pizza
> slice 1, 2, or 3 ... whichever one the mental flip lands on ... the OTHER
> two are fixed choices and out of the control of the S ... 
> 
> some of the choices DEPEND on what has already transpired
> 
> 
> 
> At 03:00 PM 2/23/01 -0600, Mike Granaas wrote:
> >On Fri, 23 Feb 2001, dennis roberts wrote:
> >
> > 
> >> 
> >> but, what is really the p for success? q for failure?
> >> 
> >> is this situation of n=10 ... really a true binomial case where p for 
> >> success is 1/3 under the  assumption that simple guessing were the way in 
> >> which tasters made their decisions?
> >
> >It's late on friday so I could be missing something, but it seems
> >reasonable that p = 1/3 in this case.  If the taster were to simply walk
> >into the room and point at the middle piece of pizza each trial they
> >should be right 1 time in 3. (Unless there is some experimental
> >manipulation that keeps the odd piece in one position more frequently than
> >would be expected...but I think you specified counterbalancing in your
> >question.)
> >
> >> 
> >> (as an aside, what would it mean for tasters in this situation to be making 
> >> their decisions purely based on chance?)
> >
> >I would interpret it as meaning that the tasters couldn't tell the two
> >pizza brands apart.  They did no better than someone who didn't taste the
> >pizza and so were unable to discriminate between to two brands.  The
> >obivious explanations are that the pizza brands really are the same in all
> >ways that matter for tast

Re: pizza

[J. Williams]

Why test them with 3 pieces?  Why not just two pieces of each and let
them identify the two brands?  Alternate the order in which they eat
the two pieces to factor out ordering of the responses.  I always like
the first piece of pizza best, don't you?  That's when you are the
hungriest :-)  Finally, see how many correct choices the students make
and if that differs what could be expected by chance.  


On 23 Feb 2001 12:10:14 -0800, [EMAIL PROTECTED] (dennis roberts) wrote:

>let's say that you have 'students' (they love pizza you know!) who claim 
>they can easily tell the difference between brands of pizza (pizza hut, 
>dominoes, etc.) ... so, you put them up to the challenge
>
>you select 10 students at random ... and, arrange a taste test as follows:
>
>you have some piping hot pizzas ... from dominoes and pizza hut ... and, 
>you cut slices of each (pepperoni and green peppers in all cases)  and, 
>when each student comes in ... you randomly pick 2 slices from one of the 
>two brands ... and 1 from the other brand ... and lay them out in front of 
>the student in a random order and ask the student to taste test ... then 
>tell you which two of the 3 are the same ... and which 1 of the 3 is 
>different ...
>
>of course, they have to try all 3 ... and, probably go back and forth 
>retasting more than once before making their final decision ...
>
>now, we have 10 trials in terms of students doing independent tests, one 
>from the other ...
>
>in each of these 10 cases ... if the identification of the 3 is correct ... 
>you count this as a successful identification ... if there are any 
>misplacements or misidentifications ... then we label this as a failure ...
>
>say we have pizza 1, 2, and 3 ... and the only allowable options are:
>
>12 same, 3 different
>13 same, 2 different
>23 same, 1 different
>
>that is, the instructions are such that they are told ... 2 ARE the same 
>... and, 1 IS different so, saying all are the same ... or all are 
>different ... are not options that you allow for the taster
>
>so, in this scenario, there are 10 independent trials ...
>
>but, what is really the p for success? q for failure?
>
>is this situation of n=10 ... really a true binomial case where p for 
>success is 1/3 under the  assumption that simple guessing were the way in 
>which tasters made their decisions?
>
>(as an aside, what would it mean for tasters in this situation to be making 
>their decisions purely based on chance?)
>
>_
>dennis roberts, educational psychology, penn state university
>208 cedar, AC 8148632401, mailto:[EMAIL PROTECTED]
>http://roberts.ed.psu.edu/users/droberts/drober~1.htm
>
>
>
>=
>Instructions for joining and leaving this list and remarks about
>the problem of INAPPROPRIATE MESSAGES are available at
>  http://jse.stat.ncsu.edu/
>=



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Re: Type III errors

[Jay Warner]

And all this time I thought it was a tongue-in-cheek way of saying, 
"find an answer to the wrong  question."  Ah!  How thick of me.  :)

Jay

Karl L. Wuensch wrote:

> Recently there was a discussion here involving the phrase "Type III 
> errors." I noted that others have used that phrase to mean inferring 
> the incorrect direction of effect after rejecting a nondirectional 
> hypothesis, but I was unable to give references. This week I stumbled 
> across references to such use of the term. Interested parties may 
> consult the article by Leventhal and Huynh (Psychological Methods, 
> 1996, 1, 278-292). They recommend that the probability of making a 
> Type III error be subtracted from the probability of correctly 
> rejecting the nondirectional null when computing power (since it is 
> common practice to infer a direction of effect following rejection of 
> a nondirectional null.
> 
> I have posted a summary of the article at: 
> http://core.ecu.edu/psyc/wuenschk/StatHelp/Type_III.htm
> 
> +
> Karl L. Wuensch, Department of Psychology,
> East Carolina University, Greenville NC 27858-4353
> Voice: 252-328-4102 Fax: 252-328-6283
> 
>  
> [EMAIL PROTECTED] 
> 
>  
> http://core.ecu.edu/psyc/wuenschk/klw.htm
> 

-- 
Jay Warner
Principal Scientist
Warner Consulting, Inc.
 North Green Bay Road
Racine, WI 53404-1216
USA

Ph: (262) 634-9100
FAX:(262) 681-1133
email:  [EMAIL PROTECTED]
web:http://www.a2q.com

The A2Q Method (tm) -- What do you want to improve today?




=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Kolmogorov-Smirnov two sample like test for discrete distr.

[Jan]

I have two samples (size about 50) from discrete distributions.
Does anybody know some nonparametric tests to
check whether the two samples come from
the same distribution?

For continuous distributions there is Kolmogorov-Smirnov
two sample test but I need similar test for discrete case.

Jan



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Re: pizza

[Richard A. Beldin]

This is a multi-part message in MIME format.
--3BC4BC183F2E8A0B7F3DB391
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

You raise a good question. Why not just ask them to admit that they
can't tell the difference rather than forcing a choice?

--3BC4BC183F2E8A0B7F3DB391
Content-Type: text/x-vcard; charset=us-ascii;
 name="rabeldin.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Richard A. Beldin
Content-Disposition: attachment;
 filename="rabeldin.vcf"

begin:vcard 
n:Beldin;Richard
tel;home:787-255-2142
x-mozilla-html:TRUE
url:netdial.caribe.net/~rabeldin/Home.html
org:BELDIN Consulting Services
version:2.1
email;internet:[EMAIL PROTECTED]
title:Professional Statistician (retired)
adr;quoted-printable:;;PO Box 716=0D=0A;Boquerón;PR;00622;
fn:Richard A. Beldin
end:vcard

--3BC4BC183F2E8A0B7F3DB391--



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Re: Bg/Time data extrapolation?

[Herman Rubin]

In article <[EMAIL PROTECTED]>,
Jeff Goslin  <[EMAIL PROTECTED]> wrote:
>[EMAIL PROTECTED] wrote:
>> In most situations I can imagine, this (interpolating historical data to
>> find "new" training exemplars) is not a good idea.  The presumed "new"
>> exemplars are very synthetic and reflect the interpolation procedure as
>> much as they do the data.  I would think that (unless you have a very
>> good reason for doing otherwise), your model (neural network or any
>> other empirical model) would better reflect reality by using only the
>> actual historical exemplars.

>Yes, I have recognized this to be a problem, but, it is less of a problem than
>the other problems I have at the moment.  Currently, there are very few machines
>that are capable of being attached to a human being and have their blood sugars
>read and stored at the intervals I would require.  Most of them are in
>development and have not been released to the public.  The data from these
>machines is very well guarded by the developers of those machines, presumably
>for trade secret reasons.  As such, I cannot simply ask someone for their minute
>by minute data and have them give it to me.  The second problem is one of
>money.  I'm sure if I threw money at the companies, they'd give me some data to
>work with, but I don't have any money.  I'm doing this purely for personal use
>at the moment.

However, there are published studies, many obtained the 
hard way by frequent finger testing, and in many cases by
continuous IV sampling of both glucose and insulin.  This
newsgroup is not the place to ask for it; consult some
research endocrinologists for examples of data.  I do not
know if you can get minute by minute data, but there are
devices now used for physicians to get frequent (I believe
5 minute intervals) readings over a 3-day period.  It is
not quite that good, as there is, I believe, daily
calibration.  You should be able to get a few such samples.

I should also warn you that, since diabetics are being
medicated, these medications do not have a totally smooth
effect.  Even for non-diabetics, and I suggest you also
look at records for such, it is not all that smooth.  It
would not even surprise me if there are books of typical
blood glucose curves for various types of people.

Looking at examples which have already been collected
should give you a better idea of what the problem is.
Knowing something about what can and cannot be done, I
suggest you bring in a good mathematical statistician who
does not just use simple models early in your research.
-- 
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, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
[EMAIL PROTECTED] Phone: (765)494-6054   FAX: (765)494-0558


=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Re: pizza

[DJNordlund]

[EMAIL PROTECTED] wrote:

>Interesting point.  Yes, if the Ss do something other than a random guess
>the binomial model would be violated.  The question then becomes what
>would they do if they are uncertain?  I suspect that they would fall back
>on visual inspection...which piece appears to be different than the others
>(less green pepper, more browned, etc)  Such information is probably
>relevant often enough that "guessing" would be well above 1/3.
>
>Using blindfolded Ss will deal with that problem, and gets us back to
>the question that Dennis is asking.  I'm guessing that rather than going
>through some sort of a systematic process (e.g. binary decision for the
>first piece, progress to second piece only if first piece was judged
>"same".) Ss will in fact do something more like guessing.  Only they
>will condition their guesses such that if they picked slice A as different
>on the previous trial they will first consider slices B and C on the
>current trial (they will actually avoid selecting the same slice position
>on sequential trials).  Furthermore they will try to equalize the number
>of position choices they make across the experiment so that they choose
>each of A, B, and C three times and one of those a fourth time.
>
>This leads to: trials are not independent.
>
>The question remains, are the trials not independent in any way that
>matters for the purposes of binomial probabilities?  Off the top of my
>head, I'm not sure.  I am aware of, through secondary sources, studies
>where any amount of guessing about random events (eg coin tosses) results
>in a lower number of correct outcomes than simply making the same choice
>every timethat is always choosing "tails" will be superior to making
>choices between "heads" and "tails" on different trials.  
>

<<>>

Michael,

can you provide references for those studies which show that '...always
choosing "tails" will be superior to making choices between "heads" and "tails"
on different trials.'  For a "fair" coin-tossing situation, I am at a loss as
to how the expected value of proportion of correct choices could be different
from .5 using either strategy.

Dan Nordlund


=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Re: Kolmogorov-Smirnov two sample like test for discrete distr.

[Herman Rubin]

In article <[EMAIL PROTECTED]>,
Jan  <[EMAIL PROTECTED]> wrote:
>I have two samples (size about 50) from discrete distributions.
>Does anybody know some nonparametric tests to
>check whether the two samples come from
>the same distribution?

>For continuous distributions there is Kolmogorov-Smirnov
>two sample test but I need similar test for discrete case.

One can lose some power by using randomization to spread
out the points, and then use the Kolmogorov-Smirnov or
other such tests.  One cannot get non-randomized tests
which are distribution free for discrete distributions.

Without doing this, the K-S test is conservative; that
is, the actual p-value is smaller than you think.





-- 
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, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
[EMAIL PROTECTED] Phone: (765)494-6054   FAX: (765)494-0558


=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Re: Bg/Time data extrapolation?

[Herman Rubin]

In article <[EMAIL PROTECTED]>,
Jeff Goslin <[EMAIL PROTECTED]> wrote:
>Hi Steve, thanks for your reply.

>"Simon, Steve, PhD" wrote:
>> Many of us on the list are skeptical about artificial intelligence methods.
>> They have proven useful for some applications, but they are often marketed
>> as if they can substitute for consultation with a professional statistician.

>Have no fear, my application would by no means hand you or your brethren a pink
>slip.  ;)

>> Putting aside that skepticism, have you thought about how you are going to
>> measure how successful your neural net is at mimicking a human pancreas? In
>> particular, you do not seem to have data on insulin production, so how will
>> you know that the neural net pancreas produces insulin at a rate comparable
>> to the human pancreas?

>Among the data that I am using are basal and bolus amounts of insulin injected
>by the patient(roughly translated, small amounts throughout the day(basal) and a
>large amount during meals(bolus)).  This information will be used as a baseline
>to train the neural net or genetic algorithm on how much insulin produces the
>desired result.  The data that I need to have "tweaked" or "interpolated", is
>the blood sugar readings themselves.  The standard practice of an insulin
>dependant diabetic is to check blood sugars at various times during the day, and
>then apply booster shots of insulin as needed.  This is, of course, NOT the way
>the human pancreas works.  It provides a steady stream of insulin as necessary
>as the blood glucose levels in the body rise and fall throughout the day.  By
>reading a blood sugar on a minute by minute basis and then producing smaller
>dosages of insulin over a prolonged period, the pancreas operation may be
>effectively simulated.

I suggest you learn a little more.  The "steady stream"
injects insulin into fatty tissue, from which it seeps at a
more or less constant rate into the blood stream.  A
posting by an endocrinologist stated that the half-life of
insulin in the blood stream is nine minutes, so this type
of delivery is needed.  There have been some experiments
done with direct intravenous injection of insulin.

>Since I only have data for patients where they are checking their blood sugars
>at most 10 times a day, I need to get a minute by minute best guess as to the
>condition of the blood sugar, based on the actual data I have.

I would be surprised if there have not been such studies
already.  You seem to be trying to reinvent the wheel.

>The success of the neural net will be measured on how effectively the program
>remains within the 80-120 blood sugar reading deemed "normal" for a Type I
>diabetic.  Too much insulin, the blood sugar readings dip into dangerously low
>levels, a bad thing.  Too little insulin, the blood sugar readings skyrocket,
>also, a bad thing.  By comparing the output of the AI to actual readings, I will
>be able to guage how effectively the AI is at determining how much insulin to
>give, based on previous readings.

This is not achieved for "normal" people.  I doubt that
it can be done, unless very carefully designed meals are
eaten, with only low glycemic index carbohydrates, and
not even that.  Also, the liver and kidneys convert their
stores of glycogen to blood glucose on the appropriate
triggers.

>Suffice it to say, the amounts of insulin delivered do not need tweaking.  Only
>the blood sugars.

You do not know when the insulin is delivered to the
blood stream.  The fastest insulin delivered to fatty
tissue takes hours to completely enter the blood stream.
This is another variable, and it is variable.  The only
way you could know the amount of insulin is to use
more complicated tests for it; these are usually not
done, except by medical researchers.  


-- 
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, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
[EMAIL PROTECTED] Phone: (765)494-6054   FAX: (765)494-0558


=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=

Re: pizza

[Donald Burrill]

On Sat, 24 Feb 2001, Mike Granaas wrote:

> Interesting point.  Yes, if the Ss do something other than a random guess
> the binomial model would be violated.  The question then becomes what
> would they do if they are uncertain?  I suspect that they would fall back
> on visual inspection...which piece appears to be different than the others
> (less green pepper, more browned, etc)  Such information is probably
> relevant often enough that "guessing" would be well above 1/3.

So what you would then have is evidence that Ss can in fact do better 
than "chance", but you might NOT know whether that improvement is due to 
their actually being able to perform as claimed, or to some other 
factor(s) relevant to identifying the "odd pizza out":  a human-cum-pizza 
version of "Clever Hans", pehaps?

> Using blindfolded Ss will deal with that problem, and gets us back to
> the question that Dennis is asking.  I'm guessing that rather than going
> through some sort of a systematic process (e.g. binary decision for the
> first piece, progress to second piece only if first piece was judged
> "same".) 
Umm:  Logical problems here. 
 (1) How can _first_ piece be judged "same"?  Same as what? 
 (2) Why would Ss not taste all three pizzas, given the ground rules 
Dennis specified (or implied) at the outset? 

> ... Ss will in fact do something more like guessing.  Only they
> will condition their guesses such that if they picked slice A as different
> on the previous trial they will first consider slices B and C on the
> current trial (they will actually avoid selecting the same slice position
> on sequential trials). 
How did "sequential trials" get into the 
scenario?  As I read Dennis' description, each S was to taste the three 
pizzas presented (perhaps tasting each more than once, but not attacking 
a whole 'nother SET of pizzas).

> Furthermore they will try to equalize the number
> of position choices they make across the experiment so that they choose
> each of A, B, and C three times and one of those a fourth time.

This sounds as though you thought each S were going to have ten separate 
trials at identifying the "odd pizza out", with a different set of three 
pizzas each time.  I don't see how else "choosing each of A, B, and C 
three times and one of those a fourth time" could mean anything else; 
but if I've misunderstood, doubtless your reply will explain.  
However interesting such an experiment might be, it's not the experiment 
that I thought Dennis described.
 
<  snip,  the rest  >
-- Don.
 --
 Donald F. Burrill[EMAIL PROTECTED]
 348 Hyde Hall, Plymouth State College,  [EMAIL PROTECTED]
 MSC #29, Plymouth, NH 03264 (603) 535-2597
 Department of Mathematics, Boston University[EMAIL PROTECTED]
 111 Cummington Street, room 261, Boston, MA 02215   (617) 353-5288
 184 Nashua Road, Bedford, NH 03110  (603) 471-7128



=
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
  http://jse.stat.ncsu.edu/
=