Newsgroup sci.stat.consult 21808

Directory

Hi,
I'm looking for an implementation of the tests for symmetry and
linearity given by Subba Rao and Gabr (1980, Journal of Time Series
Analysis 1(1):145-158) in Matlab.  I tried to write my own but am
getting results (using the sunspots and lynx datasets) that do not
correspond to those in the paper.  Any help would be appreciated.
Hwa-Tung Ong
h.ong@qut.edu.au
Queensland University of Technology
Brisbane, Australia

In message <2.2.32.19970106023351.00690340@206.64.128.3> - Sun, 5 Jan 1997
21:33:51 -0500William Delaney  writes:
>
>We have tried to use  STATA to derive Kappa values.Intercooled STATA >can't handle more than 250 enteries of a 5 choice table to compare >inter-rater agreement. 
I generated a fake data set with 2000 observations from two observers in a 5
choice table, and even with little memory, there was no problem. Perhaps I
don't understand what you mean by 'entries'.
[snip...]
Bill Rising

Elbereth (elbereth@ibm.net) wrote:
[snip]
: 
: My question for the group is this.  Do you use unit nonresponse weighting
: in your surveys?  Would you be reluctant to use the technique with a
: response rate of 40%?
: 
Yes to both questions.  In the case of a 40% response rate, I would argue
that the results are not generalizable to the whole population -- they do
however represent the respondents.
-- 
Michael P. Cohen                       home phone   202-232-4651
1615 Q Street NW #T-1                  office phone 202-219-1917
Washington, DC 20009-6331              office fax   202-219-2061
mcohen@cpcug.org

Dear group members:
Below is a sample data set, I subjected to Analysis of Variance.  Since
I was curious to see how the Windows Version of QPRO handles the task, I
ran the test as suggested in the instructions.  The relevant section of
the result indicated that the AOV table is wrong.  The df-values are
incorrect and so is the F-value.  I could have come to the innocent
conclusion that the groups do not differ from each, or that I should
accept the null hypothesis.   The second AOV table shows the correct
result (as taken from Zar 1981).
It seems quite remarkable that the developers would have gotten this
wrong.  Most likely the programmers did not take "missing values" into
account and I guess that they were taken as 0.  This can be tested by
running the same data set with the previous "missing value" replaced by
0.  It would yield the AOV table below (although the variances are
unequal and a normality test would have failed...).
I wonder who has had a similar experience, and if there is more of those
faults in the programme.  Although this was a very easy example and it
was clear from the df-values that there was a mistake, the question is
if we should have to test any results before we trust them?
I have sent a note to the developers but have not yet received a
response.
Sample data set:
A	B	C	D
60.8	68.7	102.6	87.9
57	67.7	102.1	84.2
65	74	100.2	83.1
58.6	66.3	96.5	85.7
61.7	69.8		90.3
Analysis of Variance as shown by the QPRO 6.02 output
		SS		df	MS		F		
Between Groups	1951.61		3	650.53667	1.271
Within Groups	8184.448	16	511.528
Total		10136.058	19				
Analysis of Variance table (from Zar 1981)
		SS		df	MS		F
Between Groups	4226.35		3	1408.78		165
Within Groups	128.35  	15	8.557
Total		4354.70		18

Claude Lijoi wrote:
> 
> I want to track results of a survey. Problem is that last year's survey
> was on a five point scale and this year's is on a seven point scale. What
> is the best way to convert from one form to another for comparison? I am
> an SPSS user.
> 
> Thanks for the interest,
> 
> Claude Lijoi
> cglijoi@erols.com
Sorry to come up with a question, and not a answer. Unfortunately i´m in 
the situation that i need software for statistical analysis, as i got a 
project. I intend to start as consultant but right now I havent got 
capital for SPSS at the moment - (would buy it later). Could you sent me 
a copy on e-mail??
Jens peter

I'm writing a document on data entry, and I'd like comments on the section
explaining the advantages for using a database for data entry as opposed
to using a spreadsheet.  I'd like some critical comments from others
who have used databases for data entry.  In particular, are there other
advantages to using a database?
This is written for doctors and nurses with minimal familiarity with
computers or statistics.
Here is the draft text.  
 represents a paragraph break in html.
There are several advantages for a database.

First, databases allow you to implement quality checks in the data.
For example, one of your variables might be gender.  It might be coded
1=Male, 2=Female, 9=Unknown (though if gender is unknown, you might want
to look at the credentials of the doctor doing the examination).
With a database, you could set up data entry in that field so that
it would beep anytime you tried to enter something other than a 1, 2,
or a 9.

Another quality check found in databases is a check to insure
that there are no duplicate id numbers in data tables where id numbers
have to be unique.  It's also possible to program a database to check
for consistencies in dates.  If the birthdate is in 1994, for example,
and the examination date is in 1987, then either your data is in error
or you have an extremely effective pre-natal care program.

Second, a database is effective at integrating data coming from
a variety of sources.  For example, you might have data coming from
a laboratory, a questionnaire, and from the medical records.  A database
makes it easy to properly link the information from all three sources.
Another example of where a database is extremely useful is in a 
multi-center clinical trial.  The database offers a standard way for
data entry that helps avoid the inconsistencies that can plague
such studies.

Of course, if you have a data set so complex as to take information
from three different sources, then you should definitely consult an
expert early in the design of your study.  Databases are nice, but
they are no substitute for careful planning.

Third, a database is more effective at handling very large data sets.
Unlike a spreadsheet, the entire data set does not have to fit into
computer memory.  Of course, this is a factor only when the data set
on the order of tens of thousands of records or more. If your dataset
is smaller than this then fitting all of the data into computer
memory is unlikely to be a problem.

The only disadvantage of a database compared to a spreadsheet is that
it sometimes require a bit more time to set up.  The extra time might
be beneficial, but for a simple data entry situation, it might just as
easily be overkill.
That's it!  Any comments are greatly appreciated.
aTdHvAaNnKcSe (Thanks in advance)
Steve Simon, ssimon@cmh.edu, Standard Disclaimer.
-------------------==== Posted via Deja News ====-----------------------
      http://www.dejanews.com/     Search, Read, Post to Usenet

PROBABILITIES AND WHEELS, COVERING DESIGNS, LOTTO etc.
In previous discussions all probability calculations which were still 
open were finally solved I think (correct me if still something is
outstanding). 
Now, I made the following IMHO interessting observation: we've seen (cf.
table below) that the probability for 6/49 type game using only 1 ticket 
 for AT LEAST 3 matching numbers is 1.86375% (= 1 in 53.6551)   and
 for EXACT    3 matching numbers is 1.76504% (= 1 in 56.6559)
Ie. either playing the _same_ 1 ticket in 54 drawings, or equally simply 
playing 54 _randomly_ selected _different_ tickets in 1 drawing, should 
give AT LEAST once 3 or more correct numbers in both cases.
Now, the connection to the wheels and covering designs: There exists 
a wheel which assures _always_ at least once 3 correct numbers; it is 
built up of 168 single tickets (IMHO it's the shortest known today
for 6/49 which always guarantees a win (>= 3)).
So, the interessting question is: why should one ever play the 
168 ticket wheel and not simply 54 randomly choosen different single
tickets? IMHO mathematically spoken both cases should offer nearly 
the same assurance and probability for hitting once or more at least 
3 correct numbers. (this maybe not 100% correct, but you can imagine 
what I mean).
A further question arises: why does such a wheel have so many tickets,
whereas the probability calculations show us that on average only 
54 are needed for AT LEAST 3. (Ok, I also would accept 100 tickets 
or so, but why even more than 3 times 54 ?!)
I think this deserves further investigation and research, not only for 
the player but also for the wheel designers and researchers.
Personally, I would recommend the interessted player to go with the 
54 tickets, and recommend the wheel designers and researchers in 
Design Theory (Covering Designs etc.) to also take into consideration 
such probability calculations. Any comments? 
Here again the whole table ignoring the bonus number (see r.g.l. for
tables with the bonus nbr, or email me)
vAll=49 vSub=6 k=6: Cn(49,6)=13983816 Cn(6,6)=1
 m                     p        cumul(p)             1:p      1:cumul(p)
(=match)        (=EXACT)     (=AT LEAST)        (=EXACT)     (=AT LEAST)
------------------------------------------------------------------------
 6       0.0000000715112 0.0000000715112  13983816.00000  13983816.00000
 5       0.0000184498995 0.0000185214108     54200.83721     53991.56757
 4       0.0009686197244 0.0009871411352      1032.39690      1013.02637
 3       0.0176504038669 0.0186375450020        56.65593        53.65514
 2       0.1323780290015 0.1510155740035         7.55412         6.62183
 1       0.4130194504848 0.5640350244883         2.42119         1.77294
 0       0.4359649755117 1.0000000000000         2.29376         1.00000
Uenal Mutlu
-- Uenal Mutlu (bm373592@muenchen.org)   
   Math Research, Designs/Codes, Data Compression Algorithms, C/C++
   Loc: Istanbul/Turkey + Munich/Germany

William B. Ware (wbware@EMAIL.UNC.EDU) wrote:
: I appears to me that a randomization test would provide a pretty
: convincing backup for the t-test.  Personally, I think that the sample
: sizes indicated below are sufficiently large to justify a t-test,
: especially as the skew is positive in both groups.
I usually like the advice given by William Ware, but I think I have 
to question this one.  
 -  The query that was posed was whether the reviewer was justified
in asking for a t-test, or if the data were actually suitable for it,
when the paper had been submitted with a Mann-Whitney U test because of
doubts about the distributions.
If we were pointing to some actual sample, then one could argue 
whether the t-test will be robust for these data.  But "n=30"  is 
certainly not (IMHO) sufficient reason to presume that a t-test, etc.,
"should be" performed, which seems to be the advice here.
One of the Usenet stats-groups had some discussion this Fall about
the origin and (lack of) validity of  "n=30"  as justification
for assuming the Central Limit Theorem would hold;  I did not 
contribute to that discussion, but I thought I learned something from
it, which I was trying to make use of, here.
Perhaps I was not clear in what I posted, and I invite anyone to
improve it, but:   to make the point another way -
Regardless of sample size, there can be outliers that make the rest of
the data useless;  and, I would never DEMAND a simple t-test on
untransformed data (do Rank, or something else)  if there is serious
skewing, or problematic outliers, especially if it was bad enough that
one insists on NOT reporting means, because they are misleading.
Part of what I posted, that was cited, where I was trying to state a
general rule  -
: On Tue, 31 Dec 1996, Richard F Ulrich wrote:
: >  -- Is the distribution suitable for a t-test?  If you should be
: > comfortable in using the *means*  to describe the outcomes, and to
: > compare them, then the t-test is proper and appropriate.  If the
: > means do not give a good representation, then it is hard for mere
: > sample size to compensate, even though the t-test is rather robust
: > (especially with equal n).  If you considered the means useful to
: > present, then there is implicit support for the use of a t-test.
: >
Rich Ulrich, biostatistician                wpilib+@pitt.edu
http://www.pitt.edu/~wpilib/index.html   Univ. of Pittsburgh

Michael Cohen wrote:
> 
> Elbereth (elbereth@ibm.net) wrote:
> [snip]
> :
> : My question for the group is this.  Do you use unit nonresponse weighting
> : in your surveys?  Would you be reluctant to use the technique with a
> : response rate of 40%?
> :
> 
> Yes to both questions.  In the case of a 40% response rate, I would argue
> that the results are not generalizable to the whole population -- they do
> however represent the respondents.
I would certainly agree with this.  I would not automatically write off 
the generalizability of the sample, however.  You should compare the 
sample to whatever you know about the population to see just how far off 
you were.  With such a low response rate it would be unwise to try to 
adjust by weighting, but if you know something about how your sample 
differs from the population you may at least be able to make some 
reasonable guesses about the population.
If the intended 50% response rate had been achieved the advice would be 
no different.  
It sounds like this company did not make good use of its $300,000.  A 
smaller sample with appropriate methodology could have produced better 
data at lower cost.

William Delaney (wdelaney@DREAMSCAPE.COM) wrote:
: We have tried to use  STATA to derive Kappa values.Intercooled STATA can't
: handle more than 250 enteries of a 5 choice table to compare inter-rater
: agreement. STATA Corp has been asked for assistance but fails to reply(3
: months). Kappa values seem to be the publication preferred values in
: ophthalmology. We are dealing with 675 records independantly read by two
 -- I have read several posts, and you seem to have misused STATA
or their advice-line.  I suggest you post a bit of data, to illustrate.
However, if your raters are giving you scores on an ordered scale, then
KAPPA is certainly NOT a reliability test to use.  Also, if there are
more that just *two*  categories, your Kappas may not compare to what
you see in the literature, where there usually are just two categories.
As to rating of scores:
With a paired t-test, you are directly given a powerful test of
whether one rather is systematically higher or lower.  The Pearson
correlation (which SPSS shows you , with a paired t)  gives a direct
measure of reliability.  If you look at the variances of the two
scores, you can further notice whether the raters use the same
effective RANGE for their ratings.
Rich Ulrich, biostatistician                wpilib+@pitt.edu
http://www.pitt.edu/~wpilib/index.html   Univ. of Pittsburgh

Uenal Mutlu wrote:
> 
> PROBABILITIES AND WHEELS, COVERING DESIGNS, LOTTO etc.
> 
> In previous discussions all probability calculations which were still
> open were finally solved I think (correct me if still something is
> outstanding).
> 
> Now, I made the following IMHO interessting observation: we've seen (cf.
> table below) that the probability for 6/49 type game using only 1 ticket
>  for AT LEAST 3 matching numbers is 1.86375% (= 1 in 53.6551)   and
>  for EXACT    3 matching numbers is 1.76504% (= 1 in 56.6559)
> 
> Ie. either playing the _same_ 1 ticket in 54 drawings, or equally simply
> playing 54 _randomly_ selected _different_ tickets in 1 drawing, should
> give AT LEAST once 3 or more correct numbers in both cases.
It is likely, but not guarenteed.  You would EXPECT to have a 3-win
after playing 54 tickets, but with such a small sample, you would
still have a good chance of wildly fluctuating results.
> Now, the connection to the wheels and covering designs: There exists
> a wheel which assures _always_ at least once 3 correct numbers; it is
> built up of 168 single tickets (IMHO it's the shortest known today
> for 6/49 which always guarantees a win (>= 3)).
> 
> So, the interessting question is: why should one ever play the
> 168 ticket wheel and not simply 54 randomly choosen different single
> tickets? IMHO mathematically spoken both cases should offer nearly
> the same assurance and probability for hitting once or more at least
> 3 correct numbers. (this maybe not 100% correct, but you can imagine
> what I mean).
Because the 168-wheel is a 100% guarentee.  If you absolutely,
positively want to ensure at least a 3-win, the best way to do it is
to use this wheel.
I see what you are getting at, saying that 54 tickets would give you
a darned good chance, but the price to pay for the 100% certain 3-win
is the larger wheel.
> A further question arises: why does such a wheel have so many tickets,
> whereas the probability calculations show us that on average only
> 54 are needed for AT LEAST 3. (Ok, I also would accept 100 tickets
> or so, but why even more than 3 times 54 ?!)
Again, the price to pay for ensuring the 3-win.
> I think this deserves further investigation and research, not only for
> the player but also for the wheel designers and researchers.
Yes, a good question.  Since it is more statistical, maybe Sharkey can
help compute it.
The question being, just how likely is it to win a 3-match while
playing 54 draws.
> Personally, I would recommend the interessted player to go with the
> 54 tickets, and recommend the wheel designers and researchers in
> Design Theory (Covering Designs etc.) to also take into consideration
> such probability calculations. Any comments?
Any approach where you play more tickets helps, if you really must
play at all.
> Here again the whole table ignoring the bonus number (see r.g.l. for
> tables with the bonus nbr, or email me)
> 
> vAll=49 vSub=6 k=6: Cn(49,6)=13983816 Cn(6,6)=1
>  m                     p        cumul(p)             1:p      1:cumul(p)
> (=match)        (=EXACT)     (=AT LEAST)        (=EXACT)     (=AT LEAST)
> ------------------------------------------------------------------------
>  6       0.0000000715112 0.0000000715112  13983816.00000  13983816.00000
>  5       0.0000184498995 0.0000185214108     54200.83721     53991.56757
>  4       0.0009686197244 0.0009871411352      1032.39690      1013.02637
>  3       0.0176504038669 0.0186375450020        56.65593        53.65514
>  2       0.1323780290015 0.1510155740035         7.55412         6.62183
>  1       0.4130194504848 0.5640350244883         2.42119         1.77294
>  0       0.4359649755117 1.0000000000000         2.29376         1.00000
Interested people can check these against the odds usually printed on
play slips, since the 6-subset is the game that is really played.
> 
> Uenal Mutlu
> 
> -- Uenal Mutlu (bm373592@muenchen.org)   
>    Math Research, Designs/Codes, Data Compression Algorithms, C/C++
>    Loc: Istanbul/Turkey + Munich/Germany

Robert Jung  wrote:
>Hi,
>
>I am stuck with the following problem:
>I want to test a series of disrete numbers for randomness using the runs up and
>down test. The series sometimes consist of data like:
>
>  1 3 5 5 5 8 3 2 1 1 4 6 7 etc.
>
>That is I encounter 'runs' of identical values (like 5 and 1) leading to
>(critical) multiple ties (as in the situation of the three consecutive 5's).
>
>My questions are: is the runs up and down test able to capture situations like
>the one described above? If not what alternatives are available?
>
>Has there been any recent work on the subjects of nonparametric tests of
>randomness? (My latest reference is Bradley 1968!!!)
More recent (and already classical) references are:
Knuth, D. E. (1981). The art of computer programming: Vol. 2.
   seminumerical algorithms (2nd ed.). Reading, MA: Addison-Wesley.
Ripley, B. D. (1987). Stochastic simulation. New York: Wiley.
For their runs test, Wallis & Moore (JASA, 1941) proposed the following
for your problem of tied values: "... the phase lengths are tabulated
separately for each possible sequence of signs of differences between
tied items; and the resultant distributions are averaged, ..."
(pp. 402-403).
With kind regards,
Patrick.
__________________________________________________________________________
Patrick Onghena			   	patrick.onghena@ped.kuleuven.ac.be
Katholieke Universiteit Leuven	   
Department of Educational Sciences	 	Tel1: +32 16 32.59.54
Vesaliusstraat 2				Tel2: +32 16 32.62.01
B-3000 Leuven (Belgium)				Fax : +32 16 32.59.34
	http://www.kuleuven.ac.be/facdep/psy/eng/onderz/methped.htm
___________________________________________________________________________

Peter Baade (baade@SPIDER.HERSTON.UQ.EDU.AU) wrote:
: Hello all.
: I am comparing the people's opinions against a "gold standard". (The use of
: the term gold standard is debatable, but for the purposes of this problem, I
: will assume its validity).
: The outcome measure is a binomial variable, measuring "success" and
: "failure". The terms "sucess" and "failure" are arbitary.
: The "gold standard" says that the probability of a "success" is (for
: example) is 0.7, and the probability of a "failure" is 0.3.
  -- Please define further.  For the way that I have used it, a 
Gold Standard says that A is diseased or not;  B is diseased or not.
It does not offer a probabilistic outcome.
A Sample may have 70% diseased cases, but then another sample may have
20%.  Your language seems odd....
Also, I find it rather unsettling to think of comparing/contrasting
an "opinion" to a Gold Standard.  I mean, if you can't label it as
a "judgment" but have to call it an "opinion", are you sure that
you are in the ball park of talking about reliability?
(On the other hand, if you are not talking about reliability, why do
you want to refer to something as a Gold standard?)
I can't figure any more without more information.
Rich Ulrich, biostatistician                wpilib+@pitt.edu
http://www.pitt.edu/~wpilib/index.html   Univ. of Pittsburgh

Wow!  How did they ever achieve a response rate of 40%?  I have done mail
surveys many times and never even got close to a 40% return.  Did they
offer nickles for three cents or something?

cannella@ozemail.com.au (Eddy Cannella) wrote:
>The use of contigency tables is not appropriate in this case as the
>hypothesis does not seek to show independence of growth form to site
>but rather seeks to compare the distribution of growth forms to sites.
>
One of the interpretations of "contingency table" analysis is that of 
independence.  But another use of chi-squared analyses is to compare 
distributions.  If you have 2 independent groups with responses placed 
into 5 categories and you want to compare these proportions, the 
"Pearson's chi-square" analysis is still appropriate.  
I may be missing the point here, but the chi-squared will tell you 
something.
Now, if the categories are ordered...they seem to be...there are a lot of 
analyses that you could perform.  Proportional odds models or weighted 
least squares analyses.  You have 4 df, so you could form 4 odds ratios 
contrasting against the zero category.  You could use ridits.  You could 
also look at means, equivalent to your t-test.

Elbereth (elbereth@ibm.net) wrote:
: We contracted a survey to an outside firm.  Our contract called for a mail
: survey witrh a minimum response rate of 50%.  Our contractor achieved 40%,
: after telephone follow-up to increase response rate.  They now want to use
: a weighting scheme to reduce nonresponse bias.
  -- It seems QUITE relevant to know, did the two kinds of responses 
look the same?   That is, comparing the early people to responses from
the added-on, telephone follow-up, are they different in either their
demographics, or in their responses?  --  IF so, that makes extrapolation
much harder to believe in;  and if they are very similar, that is
useful to state, to show the absence of that kind of bias.
: (I know that 40% is not even respectable.  Unfortunately, when executive
  -- well, 40% is not necessary unexpected, depending on the nature
of the survey, even if there is some difficulty in "respecting" it
for scientific or statistical purposes.  A lot of folks do get by
with less, but it depends on what you are surveying, doesn't it?
Rich Ulrich, biostatistician                wpilib+@pitt.edu
http://www.pitt.edu/~wpilib/index.html   Univ. of Pittsburgh

On Tue, 07 Jan 1997 13:23:44 -0700, Karl Schultz  wrote:
>> So, the interessting question is: why should one ever play the
>> 168 ticket wheel and not simply 54 randomly choosen different single
>> tickets? IMHO mathematically spoken both cases should offer nearly
>> the same assurance and probability for hitting once or more at least
>> 3 correct numbers. (this maybe not 100% correct, but you can imagine
>> what I mean).
>
>Because the 168-wheel is a 100% guarentee.  If you absolutely,
>positively want to ensure at least a 3-win, the best way to do it is
>to use this wheel.
>
>I see what you are getting at, saying that 54 tickets would give you
>a darned good chance, but the price to pay for the 100% certain 3-win
>is the larger wheel.
...
>> Personally, I would recommend the interessted player to go with the
>> 54 tickets, and recommend the wheel designers and researchers in
>> Design Theory (Covering Designs etc.) to also take into consideration
>> such probability calculations. Any comments?
>
>Any approach where you play more tickets helps, if you really must
>play at all.
It would be useful if we had a simulation software which for example 
looks something like the following:
LOTSIM v k b nruns fFixedTickets fFixedDraw ...
v             = total numbers (ie. 49)
k             = nbrs per ticket (ie. 6)
b             = nbr of random tickets (>= 1)
nruns         = nbr of random drawings (simulation) (>= 1)
fFixedTickets = randomly fill tickets once OR refill each time (0/1) 
fFixedDraw    = randomly draw once and keep OR redraw each time (0/1)
...
Output:
 m  ep en  rn rp  dn dp ...
---------------------------
 k  .. ..  .. ..  .. .. .
 .  .. ..  .. ..  .. .. .
 .  .. ..  .. ..  .. .. .
 0  .. ..  .. ..  .. .. .
---------------------------
Sum: .....
m  = matching nbrs (0..k)
ep = expected theoretic probability
en = expected theoretic frequency
rn = simulated real frequency
rp = simulated real probability
dn = +/- diff frequency
dp = +/- diff probability
...
If there is already a similar publicly available program I would 
like to hear about it. I think of programming such a thing too. 
Further comments/options/ideas to include in the program are welcome.
IMHO such a program would be helpful in 'practically' answering some 
still outstanding problems like in the case of the above 54 tickets 
in question.

=======================================================================
COMPUTATIONAL FINANCE  at the  Oregon  Graduate  Institute  of
Science & Technology (OGI)
Master of Science Concentrations in
         Computer Science & Engineering (CSE)
         Electrical Engineering (EE)
         Now Reviewing MS Applications for Fall 1997!
New! Certificate Program Designed for Part-Time Students.
For more information, contact OGI Admissions at (503)690-1027 or
admissions@admin.ogi.edu, or visit our Web site at:
         http://www.cse.ogi.edu/CompFin/
=======================================================================
Computational Finance Overview:
Advances in computing technology now enable the widespread use of
sophisticated, computationally intensive analysis techniques applied
to finance and financial markets. The real-time analysis of
tick-by-tick financial market data, and the real-time management
of portfolios of thousands of securities is now sweeping the
financial industry.  This has opened up new job opportunities for
scientists, engineers, and computer science professionals in the
field of Computational Finance.
The strong demand within the financial industry for technically-
sophisticated graduates is addressed at OGI by the Master of
Science and Certificate Programs in Computational Finance.  Unlike
a standard two year MBA, the programs are directed at training
scientists, engineers, and technically oriented financial professionals
in the area of quantitative finance.
The master's programs lead to a Master of Science in Computer Science
and Engineering (CSE track) or in Electrical Engineering (EE track).
The MS programs can be completed within 12 months on a full-time
basis.  In addition, OGI has introduced a Certificate program
designed to provide professionals in engineering and finance a means
of upgrading their skills or acquiring new skills in quantitative
finance on a part-time basis.
The Computational Finance MS concentrations feature a unique
combination of courses that provides a solid foundation in finance
at a non-trivial, quantitative level, plus the essential
core knowledge and skill sets of computer science or the information
technology areas of electrical engineering.  These skills are
important for advanced analysis of markets and for the development
of state-of-the-art investment analysis, portfolio management,
trading, derivatives pricing, and risk management systems.
The MS in CSE is ideal preparation for students interested in
securing positions in information systems in the financial industry,
while the MS in EE provides rigorous training for students interested
in pursuing careers as quantitative analysts at leading-edge
financial firms.
The curriculum is strongly project-oriented, using state-of-the-art
computing facilities and live/historical data from the world's
major financial markets provided by Dow Jones Telerate. Students
are trained in the use of high-level numerical and analytical
software packages for analyzing financial data.
OGI has established itself as a leading institution in research
and education in Computational Finance.  Moreover, OGI has strong
research programs in a number of areas that are highly relevant
for work in quantitative analysis and information systems in the
financial industry.
-----------------------------------------------------------------------
Admissions
-----------------------------------------------------------------------
Applications for entrance into the Computational Finance MS programs
for Fall Quarter 1997  are currently being considered.  The deadlines
for receipt of applications are:
        January 15 (Early Decision Deadline, decisions by February 15)
        March 15   (Final Deadline, decisions by April 15)
A candidate must hold a bachelor's degree in computer science,
engineering, mathematics, statistics, one of the biological or
physical sciences, finance, econometrics, or one of the quantitative
social sciences.  Candidates who hold advanced degrees in these
fields or who have experience in the financial industry are also
encouraged to apply.
Applications for the Certificate Program are considered on an
ongoing basis for entrance in any quarter.
----------------------------------------------------------------------
Contact Information
----------------------------------------------------------------------
For general information and admissions materials:
    Visit our web site at:
        http://www.cse.ogi.edu/CompFin/
    or contact:
        Office of Admissions
        Oregon Graduate Institute
        P.O.Box 91000
        Portland, OR 97291-1000
        E-mail: admissions@admin.ogi.edu
        Phone:  (503)690-1027
For special inquiries:
        E-mail: compfin@cse.ogi.edu
======================================================================

I have read responses to the original post regarding a 5-point scale versus
a 7-point scale and must disagree with a majority of the posts. If the items
from both years are likert in nature, they both measure "agreement." It so
happens that in one year a 5-point scale was used to measure agreement and in
the next year a 7-point scale was used to measure agreement. In both cases,
agreement was still the thing being measured. One can argue (as some have)
that there is no way to make these scales compariable. On the other hand, one
could argue (as I do) that since the same thing (agreement) is being
measured in both years, you can indeed compare the two years in some manner.
What I suggested in a private note to the poster, is that s/he create a
35 point scale and multiply the 5-point items by 7 and the 7-point items by
5. These items are then in a common metric which would allow comparisons
between the two years based on a 35-point scale. Using this strategy, you
could indeed track changes across the two years at what I would consider, a
reasonable level. The choice of a 5-point scale or a 7-point scale was
arbitrary to begin with for, how may points should a scale of agreement have?
Byron L. Davis, Ph.D.                              byron@usi.utah.edu
      Staff Consultant for Statistics and Research Methodology
Center for High Performance Computing              University of Utah

I am looking for "user friendly" Design of Experiments software. 
Something that will generate designs and analyze them completely, with
high quality associated graphical output.  Software that I have typically
seen usually falls short in one form or another or does not focus on DOE
very well.  I am a user of S-Plus (S-DOX) and Statistica/w (for graphics),
but I need something that will suit the needs of some of my less
experienced colleagues and reports.
Thanks in advance.  
Dr. Stan Prybyla
BFGoodrich Aerospace
prybyla@research.bfg.com <- please e-mail response here in addition to
posting.

Steve,
I think you covered all the main points rather well.  One point that you may
 want to add is that current databases
(such as Access, probably true of others) can also be used as a spreadsheet, so
 they are very simple to use
even for the small quick and dirty jobs.
Frank Ivis - fivis@arf.org
>>> Steve Simon  01/07/97 10:21am >>>
I'm writing a document on data entry, and I'd like comments on the section
 explaining the advantages for using a
database for data entry as opposed to using a spreadsheet.  I'd like some
 critical comments from others who have
used databases for data entry.  In particular, are there other advantages to
 using a database?
This is written for doctors and nurses with minimal familiarity with computers
 or statistics.
Here is the draft text.  
 represents a paragraph break in html.
There are several advantages for a database.

First, databases allow you to implement quality checks in the data.
For example, one of your variables might be gender.  It might be coded
1=Male, 2=Female, 9=Unknown (though if gender is unknown, you might want to look
 at the credentials of the
doctor doing the examination).
With a database, you could set up data entry in that field so that it would beep
 anytime you tried to enter something
other than a 1, 2, or a 9.

Another quality check found in databases is a check to insure that there are no
 duplicate id numbers in data tables
where id numbers have to be unique.  It's also possible to program a database to
 check for consistencies in dates.
If the birthdate is in 1994, for example, and the examination date is in 1987,
 then either your data is in error or you
have an extremely effective pre-natal care program.

Second, a database is effective at integrating data coming from a variety of
 sources.  For example, you might have
data coming from a laboratory, a questionnaire, and from the medical records.  A
 database makes it easy to
properly link the information from all three sources.
Another example of where a database is extremely useful is in a multi-center
 clinical trial.  The database offers a
standard way for data entry that helps avoid the inconsistencies that can plague
 such studies.

Of course, if you have a data set so complex as to take information from three
 different sources, then you should
definitely consult an expert early in the design of your study.  Databases are
 nice, but they are no substitute for
careful planning.

Third, a database is more effective at handling very large data sets.
Unlike a spreadsheet, the entire data set does not have to fit into computer
 memory.  Of course, this is a factor only
when the data set on the order of tens of thousands of records or more. If your
 dataset is smaller than this then
fitting all of the data into computer memory is unlikely to be a problem.

The only disadvantage of a database compared to a spreadsheet is that it
 sometimes require a bit more time to set
up.  The extra time might be beneficial, but for a simple data entry situation,
 it might just as easily be overkill.
That's it!  Any comments are greatly appreciated.
aTdHvAaNnKcSe (Thanks in advance)
Steve Simon, ssimon@cmh.edu, Standard Disclaimer.
-------------------==== Posted via Deja News ====-----------------------
      http://www.dejanews.com/     Search, Read, Post to Usenet

On 7 Jan 1997 05:24:45 GMT, lthompso@s.psych.uiuc.edu (Laura Thompson)
wrote:
>
>
>What has replaced the /method subcommand in spss manova?  I would like to
>get sequential SS, but cannot..the default is unique. I do not have access
>to version 7.0 and the glm procedure, and must use spss for this .. what
>can I do?
I don't know whether I've misunderstood the query, but according to my
copy of SPSS for Windows Release 6.0, you can still specify
/method=unique or /method=sequential in the manova procedure. Unique
is, as you say, the default.
Jay Weedon.