Hi:

This post concerns a published article, and I have uploaded the
file "consumer veterinary service expenditures.pdf" into the file
section.

I joined this group to learn more about statistics, especially as
related to animal welfare.  I am promoting a spay/neuter effort in
low-income neighborhoods, and looking for articles that could be
used to support our work.   I found this article, but I have heard
some rather generalized criticism of it (see below) and I would like
to have a better more knowledgable answer.

My questions are along the lines of ...

I would like to know what valid criticisms of this article there
might be from a statistical point of view.  Are these appropriate
and meaningful statistical treatments of such data?  Is there a
better way?  How much confidence I can place in these numbers and
conclusions?  It states that they "controlled for income" (is that a
fairly standardized calculation?) and if so the attached article
indicates that people who identified themselves as black, Native
American, or Asian are unlikely to have pet-related expenditures
even if they increase their income.  It indicates that less people
overall are seeking veterianry care each year - is that a logical
conclusion from this statistical treatment?  I have read about the
CEX data they used, and it seems like a reasonably representative
set of data.

Thank you for your time, and I appreciate your assistance with my
learning...

dan

Note:  The following was an example of the feedback I got.  I am not
sure of some of what he is trying to say.

"Well, my stats teacher thinks it is junk, "soft science", there is
not enough detail about their methods, etc.  Concerning the numbers,
they do not make it clear whether that positive correlation you
cited  was between amount or whether money was spent. From what they
suggested throughout the article about some people not spending any
money and those who do spend money spending much, how can the
correlation/marginal effect between income and spending be relevant
at more than one point on the curve? (aka the derivative or whatever
is a curve rather than a line.) How do they correct for income? How
representative were there data sets in reality? If there were
factors with negative marginal effect including rural/urban,
geography etc, shouldn't they have observed some effect when Native
Americans were in a different column than blacks?  I saw
the 'corrected for income' piece, but they don't justify their
correction."

Can anybody help me by providing the necessary information about
Duncan's multiple range test(DMRT).

hi, everybody. I hv some lotus files. Unable to open in EXCEL OR MINITAB. Can
anyone help me?

Muhammad Ibrahim Shamsi wrote:
>             no no. when you roll a die single time the sample space would
definitly be {1,2,3,4,5,6} . but the question is that roll a die until 3
appears. construct a sample space for this experiment.   ----- Original Message
---- From: Sam Sahil <samsahil2001@ yahoo.co. in> To: Statisticians_
group@yahoogroup s.co.in Sent: Thursday, August 28, 2008 10:18:17 Subject: Re:
[Statisticians_ group] need help
>             hi
>  what do you mean by sample space
>  according to my knowladge sample space will be {1,2,3,4,5,6}
>    SAHIL
>  ----- Original Message ---- From: Muhammad Ibrahim Shamsi <abrahamshamsi@
yahoo.com> To: statisticians_ group <statisticians_ group@yahoogroup s.co.in>
Sent: Thursday, 28 August, 2008 10:36:29 AM Subject: [Statisticians_ group] need
help
>  can any body answer me this question? an experiment consist of rolling a die
until a 3 appears, constrcut the sample space for this experiment.
>   
>  New Email addresses available on Yahoo! Get the Email name you've always
wanted on the new @ymail and @rocketmail. Hurry before someone else does!
>        Connect with friends all over the world.  Get Yahoo! India Messenger.
>         New Email names for you!
> Get the Email name you've always wanted on the new @ymail and @rocketmail.
> Hurry before someone else does!
>

hi, everybody. I hv some lotus files. Unable to open in EXCEL OR MINITAB. Can
anyone help me?

Muhammad Ibrahim Shamsi wrote:
>             no no. when you roll a die single time the sample space would
definitly be {1,2,3,4,5,6} . but the question is that roll a die until 3
appears. construct a sample space for this experiment.   ----- Original Message
---- From: Sam Sahil <samsahil2001@ yahoo.co. in> To: Statisticians_
group@yahoogroup s.co.in Sent: Thursday, August 28, 2008 10:18:17 Subject: Re:
[Statisticians_ group] need help
>             hi
>  what do you mean by sample space
>  according to my knowladge sample space will be {1,2,3,4,5,6}
>    SAHIL
>  ----- Original Message ---- From: Muhammad Ibrahim Shamsi <abrahamshamsi@
yahoo.com> To: statisticians_ group <statisticians_ group@yahoogroup s.co.in>
Sent: Thursday, 28 August, 2008 10:36:29 AM Subject: [Statisticians_ group] need
help
>  can any body answer me this question? an experiment consist of rolling a die
until a 3 appears, constrcut the sample space for this experiment.
>   
>  New Email addresses available on Yahoo! Get the Email name you've always
wanted on the new @ymail and @rocketmail. Hurry before someone else does!
>        Connect with friends all over the world.  Get Yahoo! India Messenger.
>         New Email names for you!
> Get the Email name you've always wanted on the new @ymail and @rocketmail.
> Hurry before someone else does!
>

Hi,

I think this link would be helpful for you..

http://en.wikipedia.org/wiki/Multiple_correlation

---
With Regards,
Thirumalesh.k

--- In Statisticians_group@..., "nickswett"
<nickswett@...> wrote:
>
> hello everyone:
>
>   I want to establish a correlation among four variables, say,
> A,B,C,D. now I have obtained enough data for each variable. My
> question is: how to correlate them? some similar correlations from
> references seem strange to me, like A=0.956*B^(-0.32)C^(0.2341)... how
> do those number come from? I was heard there are some software there,
> given inputs and generates sort of formula like the above. but not
sure.
>  textbook prefers to have a form like A=aX1+bX2+cX3... so what's the
> difference between these two equations?
>
>  Thank you .
>  nick
>

hello everyone:

   I want to establish a correlation among four variables, say,
A,B,C,D. now I have obtained enough data for each variable. My
question is: how to correlate them? some similar correlations from
references seem strange to me, like A=0.956*B^(-0.32)C^(0.2341)... how
do those number come from? I was heard there are some software there,
given inputs and generates sort of formula like the above. but not sure.
  textbook prefers to have a form like A=aX1+bX2+cX3... so what's the
difference between these two equations?

  Thank you .
  nick

Hey Group,
I'm used to SPSS and now have an indicator to create using STATA. I'm looking for the appropriate SPSS command 'COUNT' in STATA. This is what I'm trying to do:


I have 5 questions ( q1 to q5), answered 1, 2 and 3 (Never, sometimes and often). I want to create an indicator NANA where

0: all answers are never (1) ....maximum of 5 points
1: answer sometimes (2) OR never (3) for ONLY ONE question .....
2: answer sometimes (2) OR never (3) for two, three or four questions...
3: answer sometimes (2) OR never (3) for all questions
Category 2 is therefore a residual


In SPSS I used the command 'COUNT' doing something like this


count temp1 (q1, q2, q3...q5), (2,3)
if temp1 = 0 NANA = 0
if temp1 = 1 NANA = 1
if temp 1 2 le 4 NANA = 2
if temp 1 = 5 NANA = 3
...
This is quite simple in SPSS. Does anybody have an idea of how to do this in STATA? I tried egen with count, anycount, anymatch but none of them give me what I want....HELP!

Nana


DΓ©couvrez les photos les plus intΓ©ressantes du jour.
http://www.flickr.com/explore/interesting/7days/

Hi,
     Can any one help me to give me an idea about of Duncan's Multiple
range Test in Randomized Block Design(RBD) with a clear example with
data.

Hello,
Can anyone offer any thoughts on the following...

I'm completing a hierarchical mulitple regression in the field of
psychology on spss. Here's some details:

1. One dependant variable only; well behaved interval data, and no
transformation needed.

2. A variety of predictors, and I will enter them in 3 steps,
starting with demographic control vars.

3. In the third block I will test the interaction of two predictors
to determine if one of them acts as a buffer against the effects of
the other. Lets call the buffer predictor "B".

4. In the case of this potentially buffering predictor ... (this is
where my question is) ... its made up of a scale to measure a
psychological trait, and has a good cronbach alpha value. It is
basically interval data. I've divided the scale up into two parts -
each measuring a separate factor/attribute (B1 and B2). The trouble
is ... one of these 2 subscales is not normally distributed, and so
I've needed to do a transformation on it (square it), to get it
better behaved and fulfill normality assumptions. The full scale (B)
and the other factor subscale (B2) are good and don't require any
transformation...

5. Regarding #4 above, can I run the regression 3 times, using the
following as interaction variables with another predictor: (1) full
(B) predictor scale untransformed; (2) first subscale (B1)
transformed, as described; and (3) second subscale (B2),
untransformed.

6. In this vein, I understand that when analyzing interaction
effects, it is best to center the two interacting variables before
running the analysis in order to minimize collinearity issues ... if
so, what is the best approach for my plan in #5? Center all the
interaction predictors (including B1 and B2), but also do the
transformation as described??

Please let me know if you have any questions ... hopefully it is
clear enough.

Many thanks

Melissa Douglas

we must not belive all we read: rectification about "life owner virus".
click the links below:

il ne faut pas croire tout ce qu'on entend ou on lit: rectification de
l'infos sur "life owner virus". ce n'est pas un hoax(virus). voir liens
suivants:
http://www.hoax-slayer.com/life-is-beautiful-virus-hoax.html
http://www.symantec.com/security_response/writeup.jsp?docid=2002-011511-
0444-99&tabid=2
http://us.mcafee.com/virusInfo/default.asp?id=description&virus_k=99307

VERY IMPORTANT WARNING

Please Be Extremely Careful especially if using internet mail such as
Yahoo, Hotmail, AOL and so on. This information arrived this morning
direct from both Microsoft and Norton. Please send it to everybody
you know who has access to the Internet.

You may receive an apparently harmless email with a Power Point
presentation "Life is beautiful."

If you receive it DO NOT OPEN THE FILE UNDER ANY CIRCUMSTANCES, and
delete it immediately. If you open this file, a message will appear
on your screen saying: "It is too late now, your life is no longer
beautiful." Subsequently you will LOSE EVERYTHING IN YOUR PC and the
person who sent it to you will gain access to your name, e-mail and
password. This is a new virus which started to circulate on Saturday
afternoon. AOL has already confirmed the severity, and the antivirus
software's are not capable of destroying it. The virus has been
created by a hacker who calls himself "life owner."

PLEASE SEND A COPY OF THIS EMAIL TO ALL YOUR FRIENDS and ask them to
PASS IT ON IMMEDIATELY.

Dear Martin,

Sorry to reply you late.

You are talking about 90% 2-sided confident interviles based on
percentiles. I don't think it is same as coverage interval.
Confidence and Coverage intervals are conceptually different.
Confidence interval is about population parameter, whereas coverage
interval is about the individual sample values.

Regards

--- In Statisticians_group@..., Martin Holt
<martinholt42@...> wrote:
>
> Hi Madan,
> I'm no expert, but how about calculating the total, then
calculating 10% positions (5% above, 5% below) around your median
(?) point.
> Best Regards,
> Martin
> From: Madan Gopal Kundu <madan4331@...>
> To: Statisticians_group@...
> Sent: Monday, 11 August, 2008 8:24:40 PM
> Subject: [Statisticians_group] Non-parametric 95% coverage interval
>
>
> Dear Members,
>
> I need to calculate Non-parametric 95% coverage interval for Lab
data.
> Though I know the parametric method to calculate 95% coverage
> interval, I have no idea how to calculate this coverage interval
in
> Nonparametric way. I will appreciate if anybody can help me in
this.
>
> Thanks & Regards
> Madan Gopal Kundu
>
>
>
> Send instant messages to your online friends
http://uk.messenger.yahoo.com
>

Hi,
I am new to Matlab, and am trying to solve the
following optimization problem:

This is a nonlinear least squares problem. The goal is
to derive the best-fit equations of seven straight
lines (and other standard output e.g. residuals etc.).
To avoid any confusion, let's call each straight line
to be a 'rod' and the simple 3D structure to be the
'nest'. The problem is further explained below:

I have seven set of (3D) data points. Each set
represents a rod, whose best-fit equation I need to
derive. Each of these seven rods has a common
intersection (and at this intersection both rods end),
so a total of six intersections. Thus, the seven rods
together form my cube-like 'nest'.

I have tried to formulate the problem as follows
(please view the link
http://sites.google.com/site/niazarifin/matlab-project-images

if the following cannot be read):

Let,
i denote my rod number, 1 <= i <= 7
m denote the total number of points in each of the 7
point-sets (which is the same for all 7),
p denote a point in the i-th point-set, 1 <= p <= m
(x0i, y0i, z0i) denote any point on rod i
Thus, (xpi, ypi, zpi) denote the p-th point on rod i

Then, equation of rodi is:
				 (x – x0i) / ai = (y – y0i) / bi =(z – z0i) / zi
... [1]

For least squares fit, the usual (unconstrained)
equation, fi(.), is:

f(x0i, y0i, z0i, ai,  bi, ci | 1 <= i <= 7) =
&#8721;  &#8721;  {  [ bi (z0i – zpi) – ci (y0i – ypi)
]2 +
1 <= i <= 7  1 <= p <= m 	 [ ci (x0i – xpi) – ai (z0i
– zpi) ]2 +
					 [ ai (y0i – ypi) – bi (x0i – xpi) ]2 }
										  ... [2]
The distance cosine function for the i-th rod, hi(.),
1 <= i <= 7, is:
				 ai2 + bi2 + ci2 = 1 			 ... [3]

The intersection constraint for rodi & rodi+1, gi(.),
1 <= i <= 6, is:
ci [ai+1(y0i+1 – y0i) – bi+1(x0i+1 – x0i)] – ci+1
[ai(y0i+1 – y0i) – bi(x0i+1 – x0i)] – (ai+1bi –
aibi+1) (z0i+1 – z0i) = 0
													 ... [4]
([4] is obtained by using parametric equations of rodi
& rodi+1 and then equating the corresponding
coefficients.)

Combining all these, and defining the Lagrangian
&#955;i and &#956;i, we get the Jacobian:

J = f(.) + h(.) + g(.)
= f(x0i, y0i, z0i, ai,  bi, ci | 1 <= i <= 7)
+ &#8721; &#955;i (ai2 + bi2 + ci2 – 1)
1 <= i <= 7
+ &#8721; &#956;i (ci [ai+1(y0i+1 – y0i) – bi+1(x0i+1
– x0i)] – ci+1 [ai(y0i+1 – y0i) – bi(x0i+1 – x0i)] –
(ai+1bi – aibi+1) (z0i+1 – z0i))
1 <= i <= 6

Since I never used any optimizer before, I'd truly
appreciate if anybody can help on:
1.which Matlab function to use for the problem
2.what form of output I should expect
3.I know there is extensive documentation for each
Matlab function, yet I'd welcome the formulation of
this problem to directly feed to the Matlab function
chosen in 1.

Thanks for your patience! Any help would be greatly appreciated.


=======
Thanks,
S. M. Niaz Arifin
Graduate Student
Department of Computer Science and Engineering
University of Notre Dame, IN 46556, USA
Cell Phone: 469-826-4798
Website: http://smniazarifin.tripod.com

Not sure exactly what you mean by the 95% _coverage_ interval, but if
you mean a 95% confidence interval for the mean, then why not
bootstrap it from the data?

If you mean the interval in which we would expect 95% of the data to
lie, then just take the 2.5 percentile and 97.5 percentile.

Hope this helps

Giles

--- In Statisticians_group@..., "Madan Gopal Kundu"
<madan4331@...> wrote:
>
> Dear Members,
>
> I need to calculate Non-parametric 95% coverage interval for Lab data.
> Though I know the parametric method to calculate 95% coverage
> interval, I have no idea how to calculate this coverage interval in
> Nonparametric way. I will appreciate if anybody can help me in this.
>
> Thanks & Regards
> Madan Gopal Kundu
>