Dear all,

   We will publish eight theme articles on Algebraic Statistics and
   Computational Biology in the Oct. 2007 issue of Statistica
   Sinica. There will be spare copies of this issue available for
   anyone who is interested in or working on the topic. Please email
   me (karen AT stat.sinica.edu.tw) your mailing address if you want
   a copy of the Oct. issue. The theme articles on EM and MCMC
   methods in the July issue can be accessed at
   http://www.stat.sinica.edu.tw/statistica/.

   With all the best,
   Karen Li, on behalf of the co-editors
   Editorial Assistant

   -----------------------------------------------------------------
   Statistica Sinica publishes significant and original articles that
   promote the principled use of statistics along with related theory
   and methods in quantitative studies, essential to modern
   technologies and sciences.

Hi every body

I am Anadya Tripathi working with TCS.
TCS is going to recrute statistician as well as SAS programer in next
month.

So if any body instrested kindely send their resume to me at

anadya.tripathi@...

the location will be noida.

Anadya Tripathi

Hi every body

I am Anadya Tripathi working with TCS.
TCS is going to recrute statistician as well as SAS programer in next
month.

So if any body instrested kindely send their resume to me at

anadya.tripathi@...

the location will be noida.

Anadya Tripathi

Hi Members,

I understand the difference between Permuation test and Bootstrap
test. But I don't have idea about Randomization test. Please help me
out to distinguish Randomization test from Permutation test.

Thanks in advance.

Regards
Madan

Hi all

I m anadya tripathi working in TCS.
Tcs is going to recurit statistician in next comming month.
so if any body instrested kindely send their resume to me at

at anadya.tripathi@...

thanks and regarda
Anadya

Hello,

My name is Tracy Wolfe, Vice-President of Zingaro & Company
(www.zingaro.com), an executive search consulting firm specializing
in the search and selection of executives for the healthcare
industry.

I am currently working on an exciting assignment with my client,
Reliance Clinical Research Services (RCRS), a contract research
organization set up by the Reliance Group aimed at providing world-
class clinical research services to pharmaceutical, biotechnology
and medical device companies. The Reliance Group is headquartered
in Navi Mumbai, India and is recognized as India's most prestigious
business.

We are looking for dynamic individuals interested in relocating back
to India with a minimum of a Master's degree, PhD preferred and 1 -
5 years of experience in Statistics, Biostatistics or closely
related field. Additionally, we are looking for candidates with SAS
programming experience. For your review, I have attached several job
descriptions.

I would be grateful for your assistance in our efforts. If you know
a qualified candidate who might be interested, please contact me or
have them contact me directly.

Tracy Wolfe, PHR
Vice President
____________________________________________
Preferred Professional Recruiters - Solutions in Healthcare
Recruiting
a Zingaro Company
21936 Briarcliff Drive - Briarcliff, TX 78669-2012
Phone (512) 327-7275 - Fax (512) 327-1774
www.pprecruiters.com
____________________________________________

Using discriminant analysis on (say) 6 variables with 2 groups - I can
create a discrimination rule based on the loadings of the variables,
i.e. D = l1X1 + l2X2 + l3X3 + l4X4 + l5X5 + l6X6
from which I can test new data for membership of one of the two groups.

How is it possible to create a discrimination rule when there are more
than two groups - and hence more than one discriminants, e.g. for
three groups:
D1 = L11 X1 + L12 X2 + ...L16 X6
D2 = L21 X1 + L22 X2 + ...L26 X6

Any help gratefully received and also references welcome.

Peter

Hello group,

I would be very if someone could tell me how to calculate factor scores
on new cases based on the factor loadings from a PCA with Varimax
rotation made in SPSS 15.

Best regards,
Fred

Hello everyone. Does anyone know how to compute the median test ?

Thank you!

Sanjib Meitei 

Hello everyone. Does anyone know how to compute the median test ?

Thank you!

Dear all,

I am sending you the table of contents for the July 2007 issue of
Statistica Sinica. In this issue, we are featuring nine articles on
the theme topic of "EM and MCMC: Workhorses for Scientific
Computing", with some personal stories and views by Professor Xiao-
Li Meng on EM and MCMC in the "highlights". " For accessing the
articles, please click on the current issue at
http://www.stat.sinica.edu.tw/statistica/ if you are a subscriber,
or click on the "membership only area" at http://www.icsa.org/ if
you are an ICSA member.

With Best Wishes,
Karen Li---on behalf of the Co-Editors
Editorial Assistant
Statistica Sinica
---------------------------------------------------------------

EDITOR'S MELANGE

Highlights
Thirty years of EM and much more
Xiao-Li Meng

EM AND MCMC: WORKHORSES FOR SCIENTIFIC COMPUTING

A fast algorithm for the nonparametric maximum likelihood estimate
in the Cox-Gene model
I-Shou Chang, Chi-Chung Wen, Yuh-Jenn Wu and Che-Chi Yang

Sequential Monte Carlo Methods for permutation tests on truncated
data
Yuguo Chen and Jun S. Liu

A modified EM-algorithm for estimating the parameters of inverse
Gaussian distribution based on time-censored Wiener degradation data
Ming-Yung Lee and Jen Tang

Pool size selection for the sampling/importance resampling algorithm
Kim-Hung Li

Finite mixture modelling using the skew normal distribution
Tsung I. Lin, Jack C. Lee and Shu Y. Yen

Maximum likelihood inference in robust linear mixed-effects models
using multivariate t distributions
Peter X.-K. Song, Peng Zhang and Annie Qu

A fast EM algorithm for quadratic optimization subject to convex
constraints
Ming Tan, Guo-Liang Tian, Hong-Bin Fang and Kai Wang Ng

Functional principal component analysis for longitudinal and
survival data
Fang Yao

Lookahead and piloting strategies for variable selection
Junni L. Zhang, Ming T. Lin, Jun S. Liu and Rong Chen

GENERAL

An algebraic construction of minimally-supported D-optimal designs
for weighted polynomial regression
Fu-Chuen Chang and Bo-Jung Jiang

A profile likelihood theory for the correlated gamma-frailty model
with current status family data
I-Shou Chang, Chi-Chung Wen and Yuh-Jenn Wu

Asymptotic normality under two-phase sampling designs
Jiahua Chen and J. N. K. Rao

Adaptive estimation in a nonparametric regression model with errors-
in-variables
Fabienne Comte and Marie-Luce Taupin

The maximum likelihood method with estimated nuisance parameters in
hazard rate models with discontinuities
Dietmar Ferger

Testing for the equality of k regression curves
Juan Carlos Pardo-Fernandez, Ingrid Van Keilegom and
Wenceslao Gonzalez-Manteiga

CUSUM and EWMA multi-charts for detecting a range of mean shifts
Dong Han, Fugee Tsung, Xijian Hu and Kaibo Wang

Inference under peakedness restrictions
Javier Rojo, Jose Batun-Cutz and Ramon Durazo-Arvizu

Two-level nonregular designs from quaternary linear codes
Hongquan Xu and Alan Wong

Population size estimation in proportional trapping removal and
recapture models with known capture times
Paul S. F. Yip, Liqun Xi and Liping Liu

Structure functions for regular s^{l-m} designs with multiple groups
of factors
Yu Zhu and C. F. J. Wu

Call for Letters of Intent

    MIM/TDR Research Capacity Strengthening Grants 2008
    Effective Use of Malaria Control Measures in Africa

         Application deadline: 31 August 2007

      http://www.who.int/tdr/grants/grants/mim.htm
<http://www.who.int/tdr/grants/grants/mim.htm>


The Multilateral Initiative on Malaria (MIM) was established in response
to the pressing need to address health problems in Africa. It is funded
by multiple donors and aims to maximize the impact of scientific
research on the reduction of morbidity and mortality of malaria in
Africa.  The Special Programme for Research and Training in Tropical
Diseases (TDR), sponsored by UNICEF / UNDP/ World Bank / WHO is
facilitating the establishment of core groups of African investigators
and institutions able to undert ake high quality research and contribute
new knowledge for effective and improved control of malaria.  The
MIM/TDR Task Force on Malaria Research Capability Strengthening in
Africa managed by the TDR promotes opportunities for collaboration
between researchers in Africa and with colleagues in other malaria
endemic and non endemic regions.  The Task force reinforces the
interaction among researchers and control managers within the Ministries
of Health.  Competitive grants awarded every year provide opportuni
ties for research partnerships, collaboration, technology transfer,
infrastructure upgrade and training within the context of the research
projects.  The overall long-term goal of the grant programme is to
establish a sustainable critical mass of human resources for malaria
research and control in Africa.


Introduction

Efforts devoted to understanding the biology (genetics, biochemistry and
behaviour) of the malaria parasite and mosquito vector in the past 2
decades have resulted in new tools and improved tools against malaria.
Drugs, insecticides and insecticide-treated bed nets (ITNs) now
constitute the main stay of the armament against malaria.  It is well
established that prompt treatment with artemisinin based combination
therapies (ACTs) prevent degeneration to severe disease and
significantly reduce the number of  deaths caused by malaria.  ITNs have
also been shown to reduce malaria episodes and mortality when used
appropriately.  Indoor residual spraying and management of insecticide
resistance reinforce the strategy of integrated vector management.
These tools must be applied in the appropriate manner for effective
results. For example, although the efficacy of ITNs is well established,
the nets must not only be available but must also be used appropriately
to be effective.  Equ ally, appropriate use of drugs by p atients or
care givers, diagnostic tests by clinicians and laboratory workers,
efficient and well-managed programmatic monitoring of case management
practices can help to prolong the effective life of ACTs by preventing
or delay the emergence of antimalarial drug resistance. It is clear that
human behaviour impacts on effective control of malaria in several ways.


The Call

MIM/TDR seeks letters of intent on research aimed at identifying,
understanding and developing strategies to overcome human behavioural
barriers to appropriate and effective use of malaria control measures.
The objective is to develop innovative approaches to encourage
behaviours that help to effectively prevent and control malaria.  This
area of research has been identified as important and presently receives
relatively limited attention in malaria research and control efforts.
Research leading to effect ive and sustainable practices in communities
(rural or urban) and among care givers are particularly encouraged.

On behalf of the MIM, TDR invites letters of intent (LOI) for
collaborative projects from research groups in Africa. The letters of
intent must focus on the research area described above as well as create
opportunities for developing African scientific and public health
leadership in research and management. Letters of intent will be
evaluated based on relevance to the call, feasibility, scientific merit.
A limited number of LOI will be selected for further development into
full proposals.  Up to 3 researc h projects will eventually be
supported.  The level of funding envisaged for each project is between
US$50,000 to US$70,000 / per year with possibility of extension based on
progress of the research project and availability of funds, for up to a
period of 3 years.


How to apply

African principal investigators working in Africa may submit LOI.
Research promoting collaborative research and training, build on
existing multidisciplinary platforms or networks, and researchers who
have previously benefited from MIM support as trainees are strongly
encouraged. The LOI should reflect partnership of PI and host
institution with other regional and international institutions.

Letters of intent should be no more than 5 A4 size pages - with a 12pt
Times New Roman printing.

The following outline is suggested:

1. Project title

2. Name and address of the proposed African principal investigator

3. A executive summary of the project (2 pages) which includes
     a. background and specific question to be addressed
     b. research and capacity building objectives of the proposed
     research
     c. expected impact of the research on the malaria situation
     d. research methods
     e. outline of capacity to be strengthened

4. Description of the research institutions and composition of the study
team

5. Estimated budget

6. Relationship with any ongoing programme, research project, network,
external funding,

7. Other research teams/institutions involved as collaborators/partners.
Contribution and benefit of each collaborating institution to the
proposed work should be clearly stated.

8. Evidence (e.g. e mail) of agreement of proposed partners  to
participate in the project (as an additional page).

9. CV of the principal investigator - no more than 1 page

Letters of intent can be submitted in English or French (submissions in
French should include a summary in English).  should be included.
Letters of intent should be sent by e- mail attachment as a single
(*.doc) file to mailto:mimtdr@... no later than 31 August 2007.

MIM/TDR will prepare the short list of letters of intent for further
development into full proposals.  Only authors of short listed letters
of intent will be notified (15th September 2007).  Full proposals must
be submitted by December 15th, 2007 and will be reviewed in March 2008.

For additional information, please contact:

Dr Olumide Ogundahunsi
Manager
MIM/TDR Task Force on Malaria Research
Capability Strengthening in Africa
TDR, World Health Organization
1211, Geneva 27
Switzerland

Tel: +41 22 791 3597
Fax: +41 22 791 4854
Email: mailto:ogundahunsio@...

--
Sharing news with the t.d.r. community, mailto:tdr-scientists@... To
unsubscribe, or subscribe, write to:  owner-tdr-scientists@... News on
UNICEF/UNDP/WorldBank/WHO-TDR  http://www.who.int/tdr/topmenu/news/
<http://www.who.int/tdr/topmenu/news/>



--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the International Public
Health Forum (IPHF)'s Listserve. To post to this group, send email to
iphf@googlegroups.com
To unsubscribe from this group, send email to iphf-unsubscribe@googlegroups.com
For more options, visit this group at http://groups.google.com/group/iphf?hl=en
-~----------~----~----~----~------~----~------~--~---

ok. thank you! but the problem is that the sample is N=20 or even
smaller. N=5. and I don't know when can I say that the distribution is
simetrical, from the parameters




--- In Statisticians_group@..., kamalakar jadhav
<kamlakar.jadhav@...> wrote:
>
>
> if ur data is coming from same popolution( ie before treatment ,
after treament)the use pair t-test oterwise use Two Sample T-test for
comparering two Mean.
>
>
> Ragards,
> Kamalesh
>
>
> To: Statisticians_group@...: paullucian_szasz@...: Fri, 3 Aug 2007
15:44:55 +0000Subject: [Statisticians_group] (unknown)
>
>
>
>
>
>
>
> one more question. how can I compute, compare data between 2 groups
in 2 situation, using chi-square. It is ok to use chi-square? the data
are in frequencies but distributions are assimetrical.
> and the problem is that in the first situation the experimental
group receives treatment and the ctrl group not. and in the second
situation, both the experimental group or ctrl group don't receive any
treatment. they just monitor something.
>
>
>
> thank you!
>
>
>
> Regards, - - - - - - - - - - - - - - Paul Lucian Szasz Department of
Computer Science Faculty of Psychology - Student Babes-Bolyai
University - Cluj-Napoca Campus Building 58-60 Teodor Mihali Street
http://www.cs.ubbcluj.ro/~paul/
> http://www.ad-astra.ro/paul
>
> don't take life too seriously, you won't get out alive!!!
>
> "Science is a willingness to accept facts even when they are opposed
to wishes." -- B. F. Skinner
>
>
> Copy addresses and emails from any email account to Yahoo! Mail -
quick, easy and free. Do it now...
>
>
>
> _________________________________________________________________
> Want to look great? Get expert opinion on beauty and skin care.
> http://content.msn.co.in/Lifestyle/AskExpert/Default01.htm
>

Hi Tracy Wolfe,
       I'm B S N Siva Kumar  presently working as a
lecturer In Statistics in DLR P.G. College (Affiliated
to Andhra University) G Mamidada ,East Godavari
District, Andhra Pradesh. I Have done my post
graduation In Statistics from Andhra University
Visakhapatnam ,Andhra Pradesh.
i here by attach my CV for your Kind perusal.

Awaiting for a Good Response.

Thanks & Regards
S N Siva Kumar Balla.
09885098342



________________________________________________________________________________\
____
Pinpoint customers who are looking for what you sell.
http://searchmarketing.yahoo.com/

Hello,

My name is Tracy Wolfe, Vice-President of Zingaro & Company
(www.zingaro.com), an executive search consulting firm specializing
in the search and selection of executives for the healthcare
industry.

I am currently working on an exciting assignment with my client,
Reliance Clinical Research Services (RCRS), a contract research
organization set up by the Reliance Group aimed at providing world-
class clinical research services to pharmaceutical, biotechnology
and medical device companies.   The Reliance Group is headquartered
in Navi Mumbai, India and is recognized as India's most prestigious
business.

We are looking for dynamic individuals interested in relocating back
to India with a minimum of a Master's degree, PhD preferred and 1 -
5 years of experience in Statistics, Biostatistics or closely
related field.  Additionally, we are looking for candidates with SAS
programming experience. For your review, I have attached several job
descriptions.

I would be grateful for your assistance in our efforts. If you know
a qualified candidate who might be interested, please contact me or
have them contact me directly.



Tracy Wolfe, PHR
Vice President
____________________________________________
Preferred Professional Recruiters - Solutions in Healthcare
Recruiting
a Zingaro Company
21936 Briarcliff Drive - Briarcliff, TX 78669-2012
Phone (512) 327-7275 - Fax (512) 327-1774
www.pprecruiters.com
____________________________________________

---

To: Statisticians_group@...
From: paullucian_szasz@...
Date: Fri, 3 Aug 2007 15:44:55 +0000
Subject: [Statisticians_group] (unknown)

one more question. how can I compute, compare data between 2 groups in 2 situation, using chi-square. It is ok to use chi-square? the data are in frequencies but distributions are assimetrical.

and the problem is that in the first situation the experimental group receives treatment and the ctrl group not. and in the second situation, both the experimental group or ctrl group don't receive any treatment. they just monitor something.

 

 

 

thank you!

 

 

 

Regards,

- - - - - - - - - - - - - -
Paul Lucian Szasz
Department of Computer Science
Faculty of Psychology - Student
Babes-Bolyai University - Cluj-Napoca
Campus Building
58-60 Teodor Mihali Street
http://www.cs.ubbcluj.ro/~paul/

http://www.ad-astra.ro/paul

 

don't take life too seriously, you won't get out alive!!!

 

"Science is a willingness to accept facts even when they are opposed to wishes." -- B. F. Skinner

---

Copy addresses and emails from any email account to Yahoo! Mail - quick, easy and free. Do it now...

Dear Shyam,

---

To: Statisticians_group@...
From: agarwalshyam2003@...
Date: Thu, 2 Aug 2007 14:47:17 +0000
Subject: [Statisticians_group] MINITAB HELP

Hii i am a student doing a research on asset pricing model as the part
of may reseach i have to regress return of 120 companys can anyone
help me how go about doing it

Hii i am a student doing a research on asset pricing model as the part
of may reseach i have to regress return of 120 companys can anyone
help me how go about doing it

NO!   NO!   NO!
Do not convert continuous data to ordinal, unless you absolutely
_must_!  (that is, only when your continuous scale really isn't very
sound.)

An ordinal scale contains less information than an interval or ratio
scale (both these are 'continuous').  The ordinal scale does not
specify the 'distance' between points, while the other two do.

Therefore, if you convert a continuous scale to an ordinal one, you
are giving away information.  The minimum sample size to draw a
conclusion just went up - sometimes way up.

Instead, I recommend that you explore this option:

Assure yourself that your ordinal scale is at least approximately
linear.  That is, 'no response' does not equal 'no preference,' and
each point progresses in one 'direction' on whatever scale/measure
you are interested in.

Convert the ordinal scale to an interval scale - convert the points
from whatever to 1, 2,3, 4....

_Assume_ that your responses are linear & uniformly interval, and
make your analysis.  (The discussion on this thread so far suggests a
correlation analysis between two continuous variables.)  See what you
get.

If someone (including yourself) is skeptical, drop back to a less
informative analytical method.  If it suggests the same conclusions,
then you are on good ground.

Drawbacks & questions:

How do we know that the respondents view the scale as interval?  Some
of them may give responses midway between two 'points.'
Further:  You report an average of your ordinal responses?  That, my
good friends, means that you treated it as an interval scale.  'Fess
up! (A slang term; sorry.  It means: confess that you shifted from
ordinal to interval scale yourself.)

Did all the respondents use the same intervals between points?
Probably not.  An ordinal scale is trying to assess a very difficult
thing to measure in the first place.  Otherwise, you wouldn't use it.

Is the statistical error induced by assuming uniform intervals less
or more than the statistical error induced by variations in the
original observer?  Arguably, less.  In local hospitals in Wisconsin,
USA they have a 'pain scale' on the wall, and you are asked to rate
your pain level.  Someone who never got hit by an automobile and
generally messed up simply doesn't recognize how high that scale can go.

Is the converted-to-interval scale distributed Normally (that is, a
Normal or Gaussian distribution)?  No.  It is bounded at each end, so
it behaves more like a binomial distribution.  See Gamma and other
options.  Again, is this error less or more than  the original
measurement error?  If need be, I can convert the measurements to
something 'closer' to a Normal - an arc-sine transform.  But in most
cases the effort does not produce different conclusions.  We can do
that later, before the report.

Good luck with your analysis,
Jay
On Aug 1, 2007, at 6:25 AM, paullucian_szasz wrote:

> Thanks for yours answers!
>
> what did you mean by pros and cons of any statistical methods?
>
> anxiety is measured as never, sometimes, often, always. it is an
> ordinal scale.
>
> the second variable measures the frequencies of a particular thought
> that ocur in 2 minutes. there are scores from 0, 1, 4 to , 76, 99.
>
> so, to compute biserial, I have to transform this scores so that they
> are measured on an ordinal scale?
>
> Thank you! Any help very much appreciated!
>
>
>
>
> --- In Statisticians_group@..., Sanjib Meitei
> <sanjibmeiteicha@...> wrote:
>>
>> Dear Dr. Khalifa Elmusharaf,
>>   thanks for sharing ur knowledge with us. Anyway, i gave a general
> reply to one of the questions.
>>   Since, on of the two variables under consideration by Khalifa
> Elmusharaf  was Anxiety , which is often expressed as dichotomous
> (High, low...........anxiety can take on any value inbetween, perhaps
> beyond, but it may be difficult to measure ), with underlying
> continuity and normality, I still think that biserial would be useful
> for finding the measure of association of the two variables.
>>
>>   One has to remember that, while measuring the association of two
> variables with different measurement levels, one variable has to be
> transformed to a new variable with measurement level same with the
> other one(preferably to lower level...nominal<ordianl<ratio) . This
> process is associated with a lot of compromise on the strength of the
> statistical test applied.
>>   So, the best solution is that the person better to to consider
> all the pros and cons of any statistical methods before taking any
> inferences.
>>
>>   regards
>>   Sanjib Meitei
>>   Biostatistician
>>
>>
>>
>>
>>
>>
>> Khalifa Elmusharaf <khalifa_elmusharaf@...> wrote:
>>             Dear paullucian szasz and Sanjib Meitei
>>
>>   Greetings,
>>
>>   Biserial correlation is used when an interval variable is
> correlated with a dichotomous variable which reflects an underlying
> continuous variable. Biserial correlation is rarely used any more,
> with polyserial/polychoric correlation now being preferred. Biserial
> correlation is not supported by SPSS but is available in SAS as a
> macro.
>>
>>
>>   Spearman's rho is the most common correlation for use with an
> ordinal and an interval variable. Rho for ranked data equals
> Pearson's r for ranked data. Note SPSS will assign the mean rank to
> tied values. The formula for Spearman's rho is:
>>
>>   rho = 1 - [(6*SUM(d2)/n(n2 - 1)]
>>
>>   where d is the difference in ranks. In SPSS, choose Analyze,
> Correlate, Bivariate; check Spearman's rho.
>>
>>
>>   Kendall's tau is another common correlation for use with an
> ordinal and an interval variable. Prior to computers, rho was
> preferred to tau due to computational ease. Now that computers have
> rendered calculation trivial, tau is generally preferred. Partial
> Kendall's tau is also available as an ordinal analog to partial
> Pearsonian correlation. In SPSS, choose Analyze, Correlate,
> Bivariate; check Kendall's tau.
>>
>>
>>   Regards,
>>
>>   Dr. Khalifa Elmusharaf
>>   Registrar OB & GYN
>>   Biostatistician
>>   Faculty of Medicine
>>   University of Medical Sciences & Technology
>>   Khartoum – Sudan
>>   www.amst-edu.com
>>
>>
>> Sanjib Meitei <sanjibmeiteicha@...> wrote:       Biserial rb would
> be appropriater for such analysis
>>
>> paullucian_szasz <paullucian_szasz@...> wrote:         Hi, everyone!
>>   How can I compute a correlation test for two set of data that are
> one set ordinal scale from an anxiety test (never, sometimes, often,
> etc) and the second set of data is in frequencies - ratio scale ?
>>   The second set o data is from 2 groups (experimental and
> control), and the distributions are asymmetrical. and I have no idea
> how to compute or what type of nonparametric test is good for my
> data. I need to compare 2 groups in first setting (variable
> manipulated or treatment) and see the diferences, and the same 2
> groups (20 subject in each group) in the second setting (no
> treatment) and see the diferences.
>>   And then test if there is a relationship of the effects obtained
> with the results from the scale. Here I want to see if or how the
> results correlate.
>>   Thank you!
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> ---------------------------------
>>   Unlimited freedom, unlimited storage. Get it now
>>
>>
>>
>>
>>
>> ---------------------------------
>>   Get the Yahoo! toolbar and be alerted to new email wherever
> you're surfing.
>>
>>
>>
>>
>> ---------------------------------
>>  Why delete messages? Unlimited storage is just a click away.
>>
>
>
>
>
> ______________________________________________________________________
> __
> NOTE: The Statisticians_group is for discussing on different issues
> related to the subject Statistics, Statistical softwares like SAS,
> SPSS etc. The users are encouraged to post jobs those are
> exclusively for Statisticians like data analyst, business analyst etc.
>
> Please don't post or send any SPAM!
>
> NOTE: Please don't use this mail for any personal discussion.
>
> ______________________________________________________________________
> __
>
>
> Yahoo! Groups Links
>
>
>
>

Jay Warner
Principal Scientist
Warner Consulting, Inc.
4444 North Green Bay Road
Racine, WI 53404-1216
USA

Ph:       262.634.9100
Fax:     262.681.1133
email:  quality@...
web:    www.a2q.com

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

Thanks for yours answers!

what did you mean by pros and cons of any statistical methods?

anxiety is measured as never, sometimes, often, always. it is an
ordinal scale.

the second variable measures the frequencies of a particular thought
that ocur in 2 minutes. there are scores from 0, 1, 4 to , 76, 99.

so, to compute biserial, I have to transform this scores so that they
are measured on an ordinal scale?

Thank you! Any help very much appreciated!




--- In Statisticians_group@..., Sanjib Meitei
<sanjibmeiteicha@...> wrote:
>
> Dear Dr. Khalifa Elmusharaf,
>   thanks for sharing ur knowledge with us. Anyway, i gave a general
reply to one of the questions.
>   Since, on of the two variables under consideration by Khalifa
Elmusharaf  was Anxiety , which is often expressed as dichotomous
(High, low...........anxiety can take on any value inbetween, perhaps
beyond, but it may be difficult to measure ), with underlying
continuity and normality, I still think that biserial would be useful
for finding the measure of association of the two variables.
>
>   One has to remember that, while measuring the association of two
variables with different measurement levels, one variable has to be
transformed to a new variable with measurement level same with the
other one(preferably to lower level...nominal<ordianl<ratio) . This
process is associated with a lot of compromise on the strength of the
statistical test applied.
>   So, the best solution is that the person better to to consider
all the pros and cons of any statistical methods before taking any
inferences.
>
>   regards
>   Sanjib Meitei
>   Biostatistician
>
>
>
>
>
>
> Khalifa Elmusharaf <khalifa_elmusharaf@...> wrote:
>             Dear paullucian szasz and Sanjib Meitei
>
>   Greetings,
>
>   Biserial correlation is used when an interval variable is
correlated with a dichotomous variable which reflects an underlying
continuous variable. Biserial correlation is rarely used any more,
with polyserial/polychoric correlation now being preferred. Biserial
correlation is not supported by SPSS but is available in SAS as a
macro.
>
>
>   Spearman's rho is the most common correlation for use with an
ordinal and an interval variable. Rho for ranked data equals
Pearson's r for ranked data. Note SPSS will assign the mean rank to
tied values. The formula for Spearman's rho is:
>
>   rho = 1 - [(6*SUM(d2)/n(n2 - 1)]
>
>   where d is the difference in ranks. In SPSS, choose Analyze,
Correlate, Bivariate; check Spearman's rho.
>
>
>   Kendall's tau is another common correlation for use with an
ordinal and an interval variable. Prior to computers, rho was
preferred to tau due to computational ease. Now that computers have
rendered calculation trivial, tau is generally preferred. Partial
Kendall's tau is also available as an ordinal analog to partial
Pearsonian correlation. In SPSS, choose Analyze, Correlate,
Bivariate; check Kendall's tau.
>
>
>   Regards,
>
>   Dr. Khalifa Elmusharaf
>   Registrar OB & GYN
>   Biostatistician
>   Faculty of Medicine
>   University of Medical Sciences & Technology
>   Khartoum – Sudan
>   www.amst-edu.com
>
>
> Sanjib Meitei <sanjibmeiteicha@...> wrote:       Biserial rb would
be appropriater for such analysis
>
> paullucian_szasz <paullucian_szasz@...> wrote:         Hi, everyone!
>   How can I compute a correlation test for two set of data that are
one set ordinal scale from an anxiety test (never, sometimes, often,
etc) and the second set of data is in frequencies - ratio scale ?
>   The second set o data is from 2 groups (experimental and
control), and the distributions are asymmetrical. and I have no idea
how to compute or what type of nonparametric test is good for my
data. I need to compare 2 groups in first setting (variable
manipulated or treatment) and see the diferences, and the same 2
groups (20 subject in each group) in the second setting (no
treatment) and see the diferences.
>   And then test if there is a relationship of the effects obtained
with the results from the scale. Here I want to see if or how the
results correlate.
>   Thank you!
>
>
>
>
>
>
>
>
>
>
>
>
> ---------------------------------
>   Unlimited freedom, unlimited storage. Get it now
>
>
>
>
>
> ---------------------------------
>   Get the Yahoo! toolbar and be alerted to new email wherever
you're surfing.
>
>
>
>
> ---------------------------------
>  Why delete messages? Unlimited storage is just a click away.
>