From JCASE@cvdls.ucdavis.edu Fri Feb 16 09:18:41 2001 Received: from mxu4.u.washington.edu (mxu4.u.washington.edu [140.142.33.8]) by lists.u.washington.edu (8.9.3+UW00.05/8.9.3+UW00.12) with ESMTP id JAA52048 for ; Fri, 16 Feb 2001 09:18:38 -0800 Received: from hathor.davis.cvdls (sphinx.ucdavis.edu [169.237.116.222]) by mxu4.u.washington.edu (8.9.3+UW00.02/8.9.3+UW99.09) with ESMTP id JAA20804 for ; Fri, 16 Feb 2001 09:18:38 -0800 Received: from dragon (dragon.davis.cvdls [192.168.1.254]) by hathor.davis.cvdls (8.9.3/8.9.3) with SMTP id JAA28889 for ; Fri, 16 Feb 2001 09:18:37 -0800 (PST) Received: from Thurman-Message_Server by dragon with Novell_GroupWise; Fri, 16 Feb 2001 09:22:33 -0800 Message-Id: X-Mailer: Novell GroupWise Internet Agent 5.5.3.1 Date: Fri, 16 Feb 2001 09:22:11 -0800 From: "James Case" To: Subject: RE: testing spatial association Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="=_772CCDC9.6D0C628A" This is a MIME message. If you are reading this text, you may want to consider changing to a mail reader or gateway that understands how to properly handle MIME multipart messages. --=_772CCDC9.6D0C628A Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable The main way to test an association is to define the hypothesis and then = run the statistical test. The hypothesis defines the nature of the test. = If you hypothesis is that there is no difference between respiratory cases = near a road from those far away (and the definition of near and far must = be made in advance), then you have a two sided test for significane. If = your hypothesis is that cases occur near a road with higher frequency than = far away from a road, thenyou have a one-sided test. You must define the = parameters of the association prior to the statistic testing for the test = to be valid. the maps give you an idea of where to start. Jim Case James T. Case DVM, PhD Professor, Clinical Diagnostic Informatics CAHFS School of Veterinary Medicine, UC Davis (530)752-4408 >>> healthmaps@home.com 02/16/01 08:57AM >>> We have some real spatial staticians on this listserve, so I'll let them respond, but how did you make the map? However, the scale needs to be specified to get any idea if there is some point spatial process going on that is significant. I am not sure I agree with your assessment that there is a relationship of cases with the yellow arrows. I'd be interested in see the rates in each of the civil divisions = on a map. You may be looking at an effect due to population density more than any connection to the roads and the big roads are in areas where the population density is highest. Also beyond a certain distance from the = road there is unlikely to be a contributory effect of the big roads to COPD, that's why the scale of the problem is important. Richard Hoskins healthmaps@home.com GMT -8 -----Original Message----- From: WAPHGIS-owner@u.washington.edu [mailto:WAPHGIS-owner@u.washington.edu]On Behalf Of AZucchi@asl.bergamo.it Sent: Friday, February 16, 2001 6:19 AM To: Waphgis@U.Washington.Edu Subject: testing spatial association > ---------- > From: AZucchi@asl.bergamo.it[SMTP:AZUCCHI@ASL.BERGAMO.IT] > Sent: Friday, February 16, 2001 6:19:21 AM > To: Waphgis@U.Washington.Edu > Subject: testing spatial association > Auto forwarded by a Rule > Dear Listers, I'd like to know which is the best way to proceed in testing an association= between 2 variables in space, i.e. a spatial distribution of chronic respiratory hospital admitted cases, and the localization of the main roads in the same space domain. How can I statistically test what it seems a strong association by eyes (these associations by eyes are dangerous...)? I attached (virus free guaranteed...) a JPEG image of real situation; fucsia dots are real cases hospitalized, and yellow arrows are the main roads of my province. I'd like to receive some statistical (and possibly related software) suggestion on this matter. Thank you Dr. Alberto Zucchi Epidemiology Unit ASL Bergamo Italy (See attached file: mappa asma_3d.jpg) --=_772CCDC9.6D0C628A Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Description: HTML
The main way to test an association is to define the hypothesis and = then=20 run the statistical test.  The hypothesis defines the nature of = the=20 test.  If you hypothesis is that there is no difference between = respiratory=20 cases near a road from those far away (and the definition of near and far = must=20 be made in advance), then you have a two sided test for significane.  = If=20 your hypothesis is that cases occur near a road with higher frequency than = far=20 away from a road, thenyou have a one-sided test.  You must define = the=20 parameters of the association prior to the statistic testing for the test = to be=20 valid.  the maps give you an idea of where to start.
 
Jim Case
 
James T. Case DVM, PhD
Professor, Clinical Diagnostic=20 Informatics
CAHFS
School of Veterinary Medicine, UC=20 Davis
(530)752-4408

>>> healthmaps@home.com 02/16/01 = 08:57AM=20 >>>
We have some real spatial staticians on this listserve, so = I'll=20 let them
respond, but how did you make the map?

However, the = scale=20 needs to be specified to get any idea if there is some
point spatial = process=20 going on that is significant. I am not sure I agree
with your assessment= that=20 there is a relationship of cases with the yellow
arrows. I'd be = interested in=20 see the rates in each of the civil divisions on
a map. You may be = looking at=20 an effect due to population density more than
any connection to the = roads and=20 the big roads are in areas where the
population density is highest. = Also=20 beyond a certain distance from the road
there is unlikely to be a=20 contributory effect of the big roads to COPD,
that's why the scale of = the=20 problem is important.

Richard Hoskins
healthmaps@home.com = <mailto:healthmaps@home.com>
G= MT=20 -8


-----Original Message-----
From:=20 WAPHGIS-owner@u.washington.edu
[mailto:WAPHGIS-owner@u.wa= shington.edu]On=20 Behalf Of
AZucchi@asl.bergamo.it
Sent: Friday, February 16, 2001 = 6:19=20 AM
To: Waphgis@U.Washington.Edu
Subject: testing spatial=20 association




> ----------
> From:   &n= bsp;=20 AZucchi@asl.bergamo.it[SMTP:AZUCCHI@ASL.BERGAMO.IT]
> Sent:=20     Friday, February 16, 2001 6:19:21 AM
> To:=20     Waphgis@U.Washington.Edu
> Subject:   &n= bsp;=20 testing spatial association
> Auto forwarded by a Rule
>
Dea= r=20 Listers,
I'd like to know which is the best way to proceed in testing = an=20 association
between 2 variables in space, i.e.
a spatial distribution= of=20 chronic respiratory hospital admitted cases, and
the localization of = the main=20 roads in the same space domain.
How can I statistically test what it = seems a=20 strong association by eyes
(these associations by eyes are dangerous...)= ? I=20 attached (virus free
guaranteed...) a JPEG image of real situation; = fucsia=20 dots are real cases
hospitalized, and yellow arrows are the main roads = of my=20 province.
I'd like to receive some statistical (and possibly related=20 software)
suggestion on this matter.

Thank you
Dr. Alberto=20 Zucchi
Epidemiology Unit
ASL Bergamo
Italy

(See attached = file:=20 mappa asma_3d.jpg)



--=_772CCDC9.6D0C628A-- .