[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Cohen data suspect at best



Kai, this is what would be called a systematic bias within the entire data 

structure. As Morgenstern has previously pointed out, these group level 

biases are unbounded an can result in inverse associations.  Another 

example:



Am J Epidemiol  1998 Sep 1;148(5):475-86



Use of census-based aggregate variables to proxy for socioeconomic group:

evidence from national samples.



Geronimus AT, Bound J.



Department of Health Behavior and Health Education and the Population 

Studies

Center, University of Michigan, Ann Arbor 48109-2029, USA.



Increasingly, investigators append census-based socioeconomic 

characteristics of residential areas to individual records to address the 

problem of inadequate

socioeconomic information on health data sets. Little empirical attention 

has been given to the validity of this approach. The authors estimate health 

outcome

equations using samples from nationally representative data sets linked to 

census data.  Aggregate variables are highly multicollinear. Associations of 

health outcomes with aggregate measures are substantially weaker than with 

microlevel measures. The authors conclude that aggregate measures can not be 

interpreted as if they  were microlevel variables nor should a specific 

aggregate measure be interpreted to

represent the effects of what it is labeled.











>From: Kai Kaletsch <info@eic.nu>

>To: Rad health <healthrad@HOTMAIL.COM>, RuthWeiner@AOL.COM, blc+@PITT.EDU, 

>radsafe@list.vanderbilt.edu

>Subject: Re: Cohen data suspect at best

>Date: Mon, 28 Jan 2002 22:31:47 -0600

>

>This example would bias the results toward the null (i.e. true negative

>correlation is bigger than that reported by Cohen), unless the number of

>drunken census takers is related to the average radon level in a county.

>

>Kai

>

>From: "Rad health" <healthrad@HOTMAIL.COM>

> >

> > Ruth, I think you are wrong.  Can you document your statement concerning

>the

> > accuracy of census information?

> >

> > Dr. Field emailed me this reference below as an example of a factor that

> > would not be linear between counties and could not be corrected for 

>using

> > Cohen's stratifications.  As Dr. Field put it to me, we know both 

>smoking

> > and lung cancer rates are higher for blacks than whites.  If you have 

>data

> > like Cohen's that under report blacks, Cohen will not be able to adjust

>for

> > it between counties and for sure not within counties.  This misreporting

>of

> > SES and smoking along with other covariates asssociated with SES is 

>enough

> > to explain the inverse association alone.  Dr. Cohen, I think this

>warrants

> > the $5000.00 don't you, please donate the money to the American Lung

> > Association.

> >

> > Don

> >

> > Annals of  Epidemiology  2001 Apr 11:171-193

> >

> > The use of census data for determining race and education as SES

>indicators:

> > a validation study.

> >

> > Kwok RK, Yankaskas BC.

> >

> > Department of Epidemiology, School of Public Health, University of North

> > Carolina at Chapel Hill, Chapel Hill, NC 27599-7515, USA.

> >

> > PURPOSE: Little research has examined the validity of using census data 

>to

> > determine an individual's socio-economic status (SES), as measured by 

>race

> > and

> > educational level. This study assessed the accuracy of using aggregate

>level

> > data from United States Census Block Groups in determining race and

> > education

> > SES indicators in a cohort of women from North Carolina. METHODS: The

>study

> > analyzed patient data from the Carolina Mammography Registry and 1990

>United

> > States Census in 21 North Carolina counties. Women (n = 39,546) were

> > geocoded to their census block group and their block group 

>characteristics

> > (surrogate

> > measures) were validated with their self-reported values on race and

> > education.

> >

> > An analysis was performed to explore whether using these surrogate

>measures

> > would affect measured associations with the self-reported values. 

>RESULTS:

> > Whites were accurately identified (84.8%) more consistently than Blacks

> > (14.1%) regardless of their urban/rural status. Women without a high

>school

> > diploma or equivalent were accurately identified (56.2%) more often than

> > those with higher education levels (45.9%). Analyses using the surrogate

> > measures were significantly different than the true values according to

> > chi-square statistics.

> >

> > CONCLUSIONS: Use of census data to derive SES indicators tends to be 

>more

> > accurate for the majority than the minority population. Researchers must

>be

> > sensitive to the ecologic fallacy when using aggregate level data such 

>as

> > the census to determine individual level characteristics.

> >

> >

> >

> > _________________________________________________________________

> > MSN Photos is the easiest way to share and print your photos:

> > http://photos.msn.com/support/worldwide.aspx

> >

> > ************************************************************************

> > You are currently subscribed to the Radsafe mailing list. To 

>unsubscribe,

> > send an e-mail to Majordomo@list.vanderbilt.edu  Put the text 

>"unsubscribe

> > radsafe" (no quote marks) in the body of the e-mail, with no subject 

>line.

>You can view the Radsafe archives at http://www.vanderbilt.edu/radsafe/

> >

>





_________________________________________________________________

Chat with friends online, try MSN Messenger: http://messenger.msn.com



************************************************************************

You are currently subscribed to the Radsafe mailing list. To unsubscribe,

send an e-mail to Majordomo@list.vanderbilt.edu  Put the text "unsubscribe

radsafe" (no quote marks) in the body of the e-mail, with no subject line. You can view the Radsafe archives at http://www.vanderbilt.edu/radsafe/