SPSSX Discussion

Methodology?

Classic

List

Threaded

12 messages Options

Samir Omerovic

Methodology?

Dear all,

I have data for approximately 20.000 women that gave birth from 2001 to 2007 in one hospital. I also
have 25 (only) women who have declared themselves as addicted to certain drugs and who gave birth
from 2001 to 2007 in the same hospital (which makes it roughly about 20.025 women in total). I would
like to test some differences in the weight of the child, blood cells, etc... between addicts and
non-addicts (between 20.000 women and 25 women). I am not sure if this is ok thing to do to use
t-test, anova, chi square, etc. with such disproportional size samples.

Any thoughts on the subject?

Samir

GfK BH
Centar za istrazivanje trzista i javnog mnijenja d.o.o.

Samir Omerovic
Istrazivac/Researcher

71000 Sarajevo, Skenderija 44
BA, Bosna i Hercegovina
Tel. +387 33 550300
Fax. +387 33 444226
www.gfk.ba
www.gfk.com

samir-9

Re: Methodology?

Hi!

I suppose looking at the differences between readings from two groups of
this kinda disproportionate samples is not worthless as long as the two
groups concerned are stable within the groups. You have mentioned that the
counter part of the 25 women were addicted to certain drugs, so you should
look into the homogeneity of their cases. If they are, this small group also
give you consistent readings whatever you are interested to look into across
the groups. Though 25 against 20,000 seems to be unparallel - we will have
to accept the scarcity of the later case ( I guess that is the situation -
not too many pregnant women come with drug addiction!)

What you can do if the question still remains is you can try to increase the
sample from 25 to 30 plus. If you can raise this to 50 to 100 - nothing like
it.

Regards,
Samir Kr Paul

----- Original Message -----
From: "Samir Omerović" <[hidden email]>
To: <[hidden email]>
Sent: 09 May, 2007 12:32 PM
Subject: Methodology?

> Dear all,
>
>
>
> I have data for approximately 20.000 women that gave birth from 2001 to
> 2007 in one hospital. I also
> have 25 (only) women who have declared themselves as addicted to certain
> drugs and who gave birth
> from 2001 to 2007 in the same hospital (which makes it roughly about
> 20.025 women in total). I would
> like to test some differences in the weight of the child, blood cells,
> etc... between addicts and
> non-addicts (between 20.000 women and 25 women). I am not sure if this is
> ok thing to do to use
> t-test, anova, chi square, etc. with such disproportional size samples.
>
> Any thoughts on the subject?
>
>
>
> Samir
>
>
>
>
>
>
>
> GfK BH
> Centar za istrazivanje trzista i javnog mnijenja d.o.o.
>
> Samir Omerovic
> Istrazivac/Researcher
>
> 71000 Sarajevo, Skenderija 44
> BA, Bosna i Hercegovina
> Tel. +387 33 550300
> Fax. +387 33 444226
> www.gfk.ba
> www.gfk.com
>
>
> --
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date: 05-May-07
> 10:34 AM
>

Melissa Ives

Re: Methodology?

In reply to this post by Samir Omerovic

You may want to look into using propensity scoring to find a comparable group of women from the 20000 to use as the comparison group. The idea is to balance groups. Briefly, you would identify characteristics of the women that would predict their likelihood of being in the addicted group of 25. Create this 'propensity for being in the addicted group' score for the 20000 women, and select a group from there that closely matches your 25 addicted women.

For more information about how this works, see:
Luellen, J.K, Shadish, W.R., Clark, M.H. (2005) Propensity Scores: An Introduction and Experimental Test. Evaluation Review 29(6): 530-558.
http://intl-erx.sagepub.com/cgi/content/abstract/29/6/530

Melissa

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of samir
Sent: Wednesday, May 09, 2007 2:07 AM
To: [hidden email]
Subject: Re: [SPSSX-L] Methodology?

Hi!

I suppose looking at the differences between readings from two groups of this kinda disproportionate samples is not worthless as long as the two groups concerned are stable within the groups. You have mentioned that the counter part of the 25 women were addicted to certain drugs, so you should look into the homogeneity of their cases. If they are, this small group also give you consistent readings whatever you are interested to look into across the groups. Though 25 against 20,000 seems to be unparallel - we will have to accept the scarcity of the later case ( I guess that is the situation - not too many pregnant women come with drug addiction!)

What you can do if the question still remains is you can try to increase the sample from 25 to 30 plus. If you can raise this to 50 to 100 - nothing like it.

Regards,
Samir Kr Paul

----- Original Message -----
From: "Samir Omerović" <[hidden email]>
To: <[hidden email]>
Sent: 09 May, 2007 12:32 PM
Subject: Methodology?

> Dear all,
>
>
>
> I have data for approximately 20.000 women that gave birth from 2001
> to
> 2007 in one hospital. I also
> have 25 (only) women who have declared themselves as addicted to
> certain drugs and who gave birth from 2001 to 2007 in the same
> hospital (which makes it roughly about
> 20.025 women in total). I would
> like to test some differences in the weight of the child, blood cells,
> etc... between addicts and non-addicts (between 20.000 women and 25
> women). I am not sure if this is ok thing to do to use t-test, anova,
> chi square, etc. with such disproportional size samples.
>
> Any thoughts on the subject?
>
>
>
> Samir
>
>
>
>
>
>
>
> GfK BH
> Centar za istrazivanje trzista i javnog mnijenja d.o.o.
>
> Samir Omerovic
> Istrazivac/Researcher
>
> 71000 Sarajevo, Skenderija 44
> BA, Bosna i Hercegovina
> Tel. +387 33 550300
> Fax. +387 33 444226
> www.gfk.ba
> www.gfk.com
>
>
> --
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date:
> 05-May-07
> 10:34 AM
>

PRIVILEGED AND CONFIDENTIAL INFORMATION
This transmittal and any attachments may contain PRIVILEGED AND
CONFIDENTIAL information and is intended only for the use of the
addressee. If you are not the designated recipient, or an employee
or agent authorized to deliver such transmittals to the designated
recipient, you are hereby notified that any dissemination,
copying or publication of this transmittal is strictly prohibited. If
you have received this transmittal in error, please notify us
immediately by replying to the sender and delete this copy from your
system. You may also call us at (309) 827-6026 for assistance.

Björn Türoque

Re: Methodology?

You know for the kinds of conclusions you want to draw, a true experiment is
always the best way to get the information. If decide to do an experiment
you could get some volunteers, (pregnant women) and randomly divide them
into two groups. Give one group the ilicit addictive substances and use the
other as the control group. You would know for certain the effects these
drugs have on the child.

I am not entierly sure, but there may be some ethical issues surrounding
this type of experiment... O shoot I'm late for my IRB meeting, got to go.

Don

On 5/9/07, Melissa Ives <[hidden email]> wrote:

>
> You may want to look into using propensity scoring to find a comparable
> group of women from the 20000 to use as the comparison group. The idea is
> to balance groups. Briefly, you would identify characteristics of the women
> that would predict their likelihood of being in the addicted group of
> 25. Create this 'propensity for being in the addicted group' score for the
> 20000 women, and select a group from there that closely matches your 25
> addicted women.
>
> For more information about how this works, see:
> Luellen, J.K, Shadish, W.R., Clark, M.H. (2005) Propensity Scores: An
> Introduction and Experimental Test. Evaluation Review 29(6): 530-558.
> http://intl-erx.sagepub.com/cgi/content/abstract/29/6/530
>
> Melissa
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
> samir
> Sent: Wednesday, May 09, 2007 2:07 AM
> To: [hidden email]
> Subject: Re: [SPSSX-L] Methodology?
>
> Hi!
>
> I suppose looking at the differences between readings from two groups of
> this kinda disproportionate samples is not worthless as long as the two
> groups concerned are stable within the groups. You have mentioned that the
> counter part of the 25 women were addicted to certain drugs, so you should
> look into the homogeneity of their cases. If they are, this small group also
> give you consistent readings whatever you are interested to look into across
> the groups. Though 25 against 20,000 seems to be unparallel - we will have
> to accept the scarcity of the later case ( I guess that is the situation -
> not too many pregnant women come with drug addiction!)
>
> What you can do if the question still remains is you can try to increase
> the sample from 25 to 30 plus. If you can raise this to 50 to 100 - nothing
> like it.
>
> Regards,
> Samir Kr Paul
>
> ----- Original Message -----
> From: "Samir Omerović" <[hidden email]>
> To: <[hidden email]>
> Sent: 09 May, 2007 12:32 PM
> Subject: Methodology?
>
>
> > Dear all,
> >
> >
> >
> > I have data for approximately 20.000 women that gave birth from 2001
> > to
> > 2007 in one hospital. I also
> > have 25 (only) women who have declared themselves as addicted to
> > certain drugs and who gave birth from 2001 to 2007 in the same
> > hospital (which makes it roughly about
> > 20.025 women in total). I would
> > like to test some differences in the weight of the child, blood cells,
> > etc... between addicts and non-addicts (between 20.000 women and 25
> > women). I am not sure if this is ok thing to do to use t-test, anova,
> > chi square, etc. with such disproportional size samples.
> >
> > Any thoughts on the subject?
> >
> >
> >
> > Samir
> >
> >
> >
> >
> >
> >
> >
> > GfK BH
> > Centar za istrazivanje trzista i javnog mnijenja d.o.o.
> >
> > Samir Omerovic
> > Istrazivac/Researcher
> >
> > 71000 Sarajevo, Skenderija 44
> > BA, Bosna i Hercegovina
> > Tel. +387 33 550300
> > Fax. +387 33 444226
> > www.gfk.ba
> > www.gfk.com
> >
> >
> > --
> > No virus found in this incoming message.
> > Checked by AVG Free Edition.
> > Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date:
> > 05-May-07
> > 10:34 AM
> >
>
>
> PRIVILEGED AND CONFIDENTIAL INFORMATION
> This transmittal and any attachments may contain PRIVILEGED AND
> CONFIDENTIAL information and is intended only for the use of the
> addressee. If you are not the designated recipient, or an employee
> or agent authorized to deliver such transmittals to the designated
> recipient, you are hereby notified that any dissemination,
> copying or publication of this transmittal is strictly prohibited. If
> you have received this transmittal in error, please notify us
> immediately by replying to the sender and delete this copy from your
> system. You may also call us at (309) 827-6026 for assistance.
>

samir-9

Re: Methodology?

Are you a doctor or something? Just wondering who is recommending such a
dreadful experiment! I am sure no expecting woman would agree to go into
such an experiment. Not only mothers wish to have healthy baby and try their
best ( some obviously can't do that - we know the reasons more or less ) to
avoid having any bad tastes during pregnancy, but some fathers do that and
care for their wives too during pregnancy! But, can we imagine the situation
if the medical researchers go the way you are suggesting keeping their
patients totally ignorant of that? And how lightly you are putting the
issues of ethics!! This seems to be like the doctors who don't bother
prescribing new medicines to their patients only for the money they get from
the pharmaceutical companies for doing this and this is being done when some
other people are arguing even animals should not be used for medical
research. What a contradiction!!

----- Original Message -----
From: "Don Asay" <[hidden email]>
To: <[hidden email]>
Sent: 09 May, 2007 11:43 PM
Subject: Re: Methodology?

> You know for the kinds of conclusions you want to draw, a true experiment
> is
> always the best way to get the information. If decide to do an experiment
> you could get some volunteers, (pregnant women) and randomly divide them
> into two groups. Give one group the ilicit addictive substances and use
> the
> other as the control group. You would know for certain the effects these
> drugs have on the child.
>
> I am not entierly sure, but there may be some ethical issues surrounding
> this type of experiment... O shoot I'm late for my IRB meeting, got to go.
>
> Don
>
>
> On 5/9/07, Melissa Ives <[hidden email]> wrote:
>>
>> You may want to look into using propensity scoring to find a comparable
>> group of women from the 20000 to use as the comparison group. The idea
>> is
>> to balance groups. Briefly, you would identify characteristics of the
>> women
>> that would predict their likelihood of being in the addicted group of
>> 25. Create this 'propensity for being in the addicted group' score for
>> the
>> 20000 women, and select a group from there that closely matches your 25
>> addicted women.
>>
>> For more information about how this works, see:
>> Luellen, J.K, Shadish, W.R., Clark, M.H. (2005) Propensity Scores: An
>> Introduction and Experimental Test. Evaluation Review 29(6): 530-558.
>> http://intl-erx.sagepub.com/cgi/content/abstract/29/6/530
>>
>> Melissa
>>
>> -----Original Message-----
>> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
>> samir
>> Sent: Wednesday, May 09, 2007 2:07 AM
>> To: [hidden email]
>> Subject: Re: [SPSSX-L] Methodology?
>>
>> Hi!
>>
>> I suppose looking at the differences between readings from two groups of
>> this kinda disproportionate samples is not worthless as long as the two
>> groups concerned are stable within the groups. You have mentioned that
>> the
>> counter part of the 25 women were addicted to certain drugs, so you
>> should
>> look into the homogeneity of their cases. If they are, this small group
>> also
>> give you consistent readings whatever you are interested to look into
>> across
>> the groups. Though 25 against 20,000 seems to be unparallel - we will
>> have
>> to accept the scarcity of the later case ( I guess that is the
>> situation -
>> not too many pregnant women come with drug addiction!)
>>
>> What you can do if the question still remains is you can try to increase
>> the sample from 25 to 30 plus. If you can raise this to 50 to 100 -
>> nothing
>> like it.
>>
>> Regards,
>> Samir Kr Paul
>>
>> ----- Original Message -----
>> From: "Samir Omerović" <[hidden email]>
>> To: <[hidden email]>
>> Sent: 09 May, 2007 12:32 PM
>> Subject: Methodology?
>>
>>
>> > Dear all,
>> >
>> >
>> >
>> > I have data for approximately 20.000 women that gave birth from 2001
>> > to
>> > 2007 in one hospital. I also
>> > have 25 (only) women who have declared themselves as addicted to
>> > certain drugs and who gave birth from 2001 to 2007 in the same
>> > hospital (which makes it roughly about
>> > 20.025 women in total). I would
>> > like to test some differences in the weight of the child, blood cells,
>> > etc... between addicts and non-addicts (between 20.000 women and 25
>> > women). I am not sure if this is ok thing to do to use t-test, anova,
>> > chi square, etc. with such disproportional size samples.
>> >
>> > Any thoughts on the subject?
>> >
>> >
>> >
>> > Samir
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> > GfK BH
>> > Centar za istrazivanje trzista i javnog mnijenja d.o.o.
>> >
>> > Samir Omerovic
>> > Istrazivac/Researcher
>> >
>> > 71000 Sarajevo, Skenderija 44
>> > BA, Bosna i Hercegovina
>> > Tel. +387 33 550300
>> > Fax. +387 33 444226
>> > www.gfk.ba
>> > www.gfk.com
>> >
>> >
>> > --
>> > No virus found in this incoming message.
>> > Checked by AVG Free Edition.
>> > Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date:
>> > 05-May-07
>> > 10:34 AM
>> >
>>
>>
>> PRIVILEGED AND CONFIDENTIAL INFORMATION
>> This transmittal and any attachments may contain PRIVILEGED AND
>> CONFIDENTIAL information and is intended only for the use of the
>> addressee. If you are not the designated recipient, or an employee
>> or agent authorized to deliver such transmittals to the designated
>> recipient, you are hereby notified that any dissemination,
>> copying or publication of this transmittal is strictly prohibited. If
>> you have received this transmittal in error, please notify us
>> immediately by replying to the sender and delete this copy from your
>> system. You may also call us at (309) 827-6026 for assistance.
>>
>

--------------------------------------------------------------------------------

No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date: 05-May-07
10:34 AM

Frederic Villamayor Forcada

Re: Methodology?

In reply to this post by Björn Türoque

What kind of joke is that?

Frederic Villamayor, BS
Researcher
Biostatistics Unit
Juan de Sada, 32
08028 Barcelona
Tel +34 935093236
Fax +34 934112764
[hidden email]
http://www.ferrergrupo.com

Este mensaje, y en su caso, cualquier fichero anexo al mismo, puede
contener información confidencial, siendo para uso exclusivo del
destinatario, quedando prohibida su divulgación, copia o distribución a
terceros sin la autorización expresa del remitente. Si Vd. ha recibido
este mensaje erróneamente, se ruega lo notifique al remitente y proceda a
su borrado. Gracias por su colaboración.

This message and its annexed files may contain confidential information
which is exclusively for the use of the addressee. It is strictly
forbidden to distribute copies to third parties without the explicit
permission of the sender. If you receive this message by mistake, please
notify it to the sender and make sure to delete it. Thank you for your
kind cooperation.

Don Asay <[hidden email]>
Enviado por: "SPSSX(r) Discussion" <[hidden email]>
09/05/2007 19:43
Por favor, responda a
Don Asay <[hidden email]>

Para
[hidden email]
cc

Asunto
Re: [SPSSX-L] Methodology?

You know for the kinds of conclusions you want to draw, a true experiment
is
always the best way to get the information. If decide to do an experiment
you could get some volunteers, (pregnant women) and randomly divide them
into two groups. Give one group the ilicit addictive substances and use
the
other as the control group. You would know for certain the effects these
drugs have on the child.

I am not entierly sure, but there may be some ethical issues surrounding
this type of experiment... O shoot I'm late for my IRB meeting, got to go.

Don

On 5/9/07, Melissa Ives <[hidden email]> wrote:
>
> You may want to look into using propensity scoring to find a comparable
> group of women from the 20000 to use as the comparison group. The idea
is
> to balance groups. Briefly, you would identify characteristics of the
women
> that would predict their likelihood of being in the addicted group of
> 25. Create this 'propensity for being in the addicted group' score for
the

> 20000 women, and select a group from there that closely matches your 25
> addicted women.
>
> For more information about how this works, see:
> Luellen, J.K, Shadish, W.R., Clark, M.H. (2005) Propensity Scores: An
> Introduction and Experimental Test. Evaluation Review 29(6): 530-558.
> http://intl-erx.sagepub.com/cgi/content/abstract/29/6/530
>
> Melissa
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
> samir
> Sent: Wednesday, May 09, 2007 2:07 AM
> To: [hidden email]
> Subject: Re: [SPSSX-L] Methodology?
>
> Hi!
>
> I suppose looking at the differences between readings from two groups of
> this kinda disproportionate samples is not worthless as long as the two
> groups concerned are stable within the groups. You have mentioned that

the
> counter part of the 25 women were addicted to certain drugs, so you
should
> look into the homogeneity of their cases. If they are, this small group
also
> give you consistent readings whatever you are interested to look into
across
> the groups. Though 25 against 20,000 seems to be unparallel - we will
have
> to accept the scarcity of the later case ( I guess that is the situation
-
> not too many pregnant women come with drug addiction!)
>
> What you can do if the question still remains is you can try to increase
> the sample from 25 to 30 plus. If you can raise this to 50 to 100 -
nothing

> like it.
>
> Regards,
> Samir Kr Paul
>
> ----- Original Message -----
> From: "Samir Omerović" <[hidden email]>
> To: <[hidden email]>
> Sent: 09 May, 2007 12:32 PM
> Subject: Methodology?
>
>
> > Dear all,
> >
> >
> >
> > I have data for approximately 20.000 women that gave birth from 2001
> > to
> > 2007 in one hospital. I also
> > have 25 (only) women who have declared themselves as addicted to
> > certain drugs and who gave birth from 2001 to 2007 in the same
> > hospital (which makes it roughly about
> > 20.025 women in total). I would
> > like to test some differences in the weight of the child, blood cells,
> > etc... between addicts and non-addicts (between 20.000 women and 25
> > women). I am not sure if this is ok thing to do to use t-test, anova,
> > chi square, etc. with such disproportional size samples.
> >
> > Any thoughts on the subject?
> >
> >
> >
> > Samir
> >
> >
> >
> >
> >
> >
> >
> > GfK BH
> > Centar za istrazivanje trzista i javnog mnijenja d.o.o.
> >
> > Samir Omerovic
> > Istrazivac/Researcher
> >
> > 71000 Sarajevo, Skenderija 44
> > BA, Bosna i Hercegovina
> > Tel. +387 33 550300
> > Fax. +387 33 444226
> > www.gfk.ba
> > www.gfk.com
> >
> >
> > --
> > No virus found in this incoming message.
> > Checked by AVG Free Edition.
> > Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date:
> > 05-May-07
> > 10:34 AM
> >
>
>
> PRIVILEGED AND CONFIDENTIAL INFORMATION
> This transmittal and any attachments may contain PRIVILEGED AND
> CONFIDENTIAL information and is intended only for the use of the
> addressee. If you are not the designated recipient, or an employee
> or agent authorized to deliver such transmittals to the designated
> recipient, you are hereby notified that any dissemination,
> copying or publication of this transmittal is strictly prohibited. If
> you have received this transmittal in error, please notify us
> immediately by replying to the sender and delete this copy from your
> system. You may also call us at (309) 827-6026 for assistance.
>

F. Gabarrot

Re: Methodology?

In reply to this post by samir-9

Hello,

You may also try to pair a non addict mother ith an addict one, as a function of a set of variables, such as age, height, weight, socio-economical status, whether they live in the city centre, or in the suburbs, etc. So that you can rule out a lot of alternative explanations. If think this will dramatically reduce the size of your non-addict mothers sample, if you choose them so they match with an addict mother. I am not able to explain this methodology further in english, but, you may find literature on this kind of randomized design in many handbooks.

Fabrice.

samir-9 wrote

Are you a doctor or something? Just wondering who is recommending such a
dreadful experiment! I am sure no expecting woman would agree to go into
such an experiment. Not only mothers wish to have healthy baby and try their
best ( some obviously can't do that - we know the reasons more or less ) to
avoid having any bad tastes during pregnancy, but some fathers do that and
care for their wives too during pregnancy! But, can we imagine the situation
if the medical researchers go the way you are suggesting keeping their
patients totally ignorant of that? And how lightly you are putting the
issues of ethics!! This seems to be like the doctors who don't bother
prescribing new medicines to their patients only for the money they get from
the pharmaceutical companies for doing this and this is being done when some
other people are arguing even animals should not be used for medical
research. What a contradiction!!

----- Original Message -----
From: "Don Asay" <donald.asay82@gmail.com>
To: <SPSSX-L@LISTSERV.UGA.EDU>
Sent: 09 May, 2007 11:43 PM
Subject: Re: Methodology?

> You know for the kinds of conclusions you want to draw, a true experiment
> is
> always the best way to get the information. If decide to do an experiment
> you could get some volunteers, (pregnant women) and randomly divide them
> into two groups. Give one group the ilicit addictive substances and use
> the
> other as the control group. You would know for certain the effects these
> drugs have on the child.
>
> I am not entierly sure, but there may be some ethical issues surrounding
> this type of experiment... O shoot I'm late for my IRB meeting, got to go.
>
> Don
>
>
> On 5/9/07, Melissa Ives <mives@chestnut.org> wrote:
>>
>> You may want to look into using propensity scoring to find a comparable
>> group of women from the 20000 to use as the comparison group. The idea
>> is
>> to balance groups. Briefly, you would identify characteristics of the
>> women
>> that would predict their likelihood of being in the addicted group of
>> 25. Create this 'propensity for being in the addicted group' score for
>> the
>> 20000 women, and select a group from there that closely matches your 25
>> addicted women.
>>
>> For more information about how this works, see:
>> Luellen, J.K, Shadish, W.R., Clark, M.H. (2005) Propensity Scores: An
>> Introduction and Experimental Test. Evaluation Review 29(6): 530-558.
>> http://intl-erx.sagepub.com/cgi/content/abstract/29/6/530
>>
>> Melissa
>>
>> -----Original Message-----
>> From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
>> samir
>> Sent: Wednesday, May 09, 2007 2:07 AM
>> To: SPSSX-L@LISTSERV.UGA.EDU
>> Subject: Re: [SPSSX-L] Methodology?
>>
>> Hi!
>>
>> I suppose looking at the differences between readings from two groups of
>> this kinda disproportionate samples is not worthless as long as the two
>> groups concerned are stable within the groups. You have mentioned that
>> the
>> counter part of the 25 women were addicted to certain drugs, so you
>> should
>> look into the homogeneity of their cases. If they are, this small group
>> also
>> give you consistent readings whatever you are interested to look into
>> across
>> the groups. Though 25 against 20,000 seems to be unparallel - we will
>> have
>> to accept the scarcity of the later case ( I guess that is the
>> situation -
>> not too many pregnant women come with drug addiction!)
>>
>> What you can do if the question still remains is you can try to increase
>> the sample from 25 to 30 plus. If you can raise this to 50 to 100 -
>> nothing
>> like it.
>>
>> Regards,
>> Samir Kr Paul
>>
>> ----- Original Message -----
>> From: "Samir Omerović" <samir.omerovic@gfk.ba>
>> To: <SPSSX-L@LISTSERV.UGA.EDU>
>> Sent: 09 May, 2007 12:32 PM
>> Subject: Methodology?
>>
>>
>> > Dear all,
>> >
>> >
>> >
>> > I have data for approximately 20.000 women that gave birth from 2001
>> > to
>> > 2007 in one hospital. I also
>> > have 25 (only) women who have declared themselves as addicted to
>> > certain drugs and who gave birth from 2001 to 2007 in the same
>> > hospital (which makes it roughly about
>> > 20.025 women in total). I would
>> > like to test some differences in the weight of the child, blood cells,
>> > etc... between addicts and non-addicts (between 20.000 women and 25
>> > women). I am not sure if this is ok thing to do to use t-test, anova,
>> > chi square, etc. with such disproportional size samples.
>> >
>> > Any thoughts on the subject?
>> >
>> >
>> >
>> > Samir
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> > GfK BH
>> > Centar za istrazivanje trzista i javnog mnijenja d.o.o.
>> >
>> > Samir Omerovic
>> > Istrazivac/Researcher
>> >
>> > 71000 Sarajevo, Skenderija 44
>> > BA, Bosna i Hercegovina
>> > Tel. +387 33 550300
>> > Fax. +387 33 444226
>> > www.gfk.ba
>> > www.gfk.com
>> >
>> >
>> > --
>> > No virus found in this incoming message.
>> > Checked by AVG Free Edition.
>> > Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date:
>> > 05-May-07
>> > 10:34 AM
>> >
>>
>>
>> PRIVILEGED AND CONFIDENTIAL INFORMATION
>> This transmittal and any attachments may contain PRIVILEGED AND
>> CONFIDENTIAL information and is intended only for the use of the
>> addressee. If you are not the designated recipient, or an employee
>> or agent authorized to deliver such transmittals to the designated
>> recipient, you are hereby notified that any dissemination,
>> copying or publication of this transmittal is strictly prohibited. If
>> you have received this transmittal in error, please notify us
>> immediately by replying to the sender and delete this copy from your
>> system. You may also call us at (309) 827-6026 for assistance.
>>
>

--------------------------------------------------------------------------------

No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date: 05-May-07
10:34 AM

Samir Omerovic

Re: Methodology?

In reply to this post by Melissa Ives

Thank you all,

This propensity scoring is a bit new to me. I will have to check it out. Thanks Melissa.

Samir, thank you too. I am still waiting for all the data so when I get it I will look if there is
homogeneity of the cases and then run the statistics and everything. I do not believe that I will
have more cases of addicts. 25 if as high as it goes.

Samir

GfK BH
Centar za istrazivanje trzista i javnog mnijenja d.o.o.

Samir Omerovic
Istrazivac/Researcher

tel. +387 33 550300
fax. +387 33 444226
www.gfk.ba
www.gfk.com
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Melissa Ives
Sent: Wednesday, May 09, 2007 5:51 PM
To: [hidden email]
Subject: Re: Methodology?

You may want to look into using propensity scoring to find a comparable group of women from the
20000 to use as the comparison group. The idea is to balance groups. Briefly, you would identify
characteristics of the women that would predict their likelihood of being in the addicted group of
25. Create this 'propensity for being in the addicted group' score for the 20000 women, and select
a group from there that closely matches your 25 addicted women.

For more information about how this works, see:
Luellen, J.K, Shadish, W.R., Clark, M.H. (2005) Propensity Scores: An Introduction and Experimental
Test. Evaluation Review 29(6): 530-558.
http://intl-erx.sagepub.com/cgi/content/abstract/29/6/530

Melissa

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of samir
Sent: Wednesday, May 09, 2007 2:07 AM
To: [hidden email]
Subject: Re: [SPSSX-L] Methodology?

Hi!

I suppose looking at the differences between readings from two groups of this kinda disproportionate
samples is not worthless as long as the two groups concerned are stable within the groups. You have
mentioned that the counter part of the 25 women were addicted to certain drugs, so you should look
into the homogeneity of their cases. If they are, this small group also give you consistent readings
whatever you are interested to look into across the groups. Though 25 against 20,000 seems to be
unparallel - we will have to accept the scarcity of the later case ( I guess that is the situation -
not too many pregnant women come with drug addiction!)

What you can do if the question still remains is you can try to increase the sample from 25 to 30
plus. If you can raise this to 50 to 100 - nothing like it.

Regards,
Samir Kr Paul

----- Original Message -----
From: "Samir Omerović" <[hidden email]>
To: <[hidden email]>
Sent: 09 May, 2007 12:32 PM
Subject: Methodology?

samir-9

Re: Methodology?

In reply to this post by Frederic Villamayor Forcada

Dear Fred:

I am sorry as I could not understand that he was actually joking and was
trying to reply him. Hope you have seen my stupidity by now.

Samir
----- Original Message -----
From: "Frederic Villamayor Forcada" <[hidden email]>
To: <[hidden email]>
Sent: 10 May, 2007 1:21 PM
Subject: Re: Methodology?

What kind of joke is that?

Frederic Villamayor, BS
Researcher
Biostatistics Unit
Juan de Sada, 32
08028 Barcelona
Tel +34 935093236
Fax +34 934112764
[hidden email]
http://www.ferrergrupo.com

Este mensaje, y en su caso, cualquier fichero anexo al mismo, puede
contener información confidencial, siendo para uso exclusivo del
destinatario, quedando prohibida su divulgación, copia o distribución a
terceros sin la autorización expresa del remitente. Si Vd. ha recibido
este mensaje erróneamente, se ruega lo notifique al remitente y proceda a
su borrado. Gracias por su colaboración.

This message and its annexed files may contain confidential information
which is exclusively for the use of the addressee. It is strictly
forbidden to distribute copies to third parties without the explicit
permission of the sender. If you receive this message by mistake, please
notify it to the sender and make sure to delete it. Thank you for your
kind cooperation.

Don Asay <[hidden email]>
Enviado por: "SPSSX(r) Discussion" <[hidden email]>
09/05/2007 19:43
Por favor, responda a
Don Asay <[hidden email]>

Para
[hidden email]
cc

Asunto
Re: [SPSSX-L] Methodology?

You know for the kinds of conclusions you want to draw, a true experiment
is
always the best way to get the information. If decide to do an experiment
you could get some volunteers, (pregnant women) and randomly divide them
into two groups. Give one group the ilicit addictive substances and use
the
other as the control group. You would know for certain the effects these
drugs have on the child.

I am not entierly sure, but there may be some ethical issues surrounding
this type of experiment... O shoot I'm late for my IRB meeting, got to go.

Don

On 5/9/07, Melissa Ives <[hidden email]> wrote:
>
> You may want to look into using propensity scoring to find a comparable
> group of women from the 20000 to use as the comparison group. The idea
is
> to balance groups. Briefly, you would identify characteristics of the
women
> that would predict their likelihood of being in the addicted group of
> 25. Create this 'propensity for being in the addicted group' score for
the

--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date: 05-May-07
10:34 AM

samir-9

Re: Methodology?

In reply to this post by Samir Omerovic

prop scoring sounds to be more methodical, but I guess you can at least
approach in line with Melissa's suggestion. Try to match a similar sample
for the addicted women that you finally come up in terms of demographics -
age, race ( if there is any) and other important socio-economic parameters
like the other person has also suggested - residence area/status, occupation
etc. This wd help increase between groups homogeneity and in case there is a
lack of it, SPSS ANOVA can address the issue with further options available
there.

All the best!

Samir

----- Original Message -----
From: "Samir Omerović" <[hidden email]>
To: <[hidden email]>
Sent: 10 May, 2007 2:35 PM
Subject: Re: Methodology?

> Thank you all,
>
> This propensity scoring is a bit new to me. I will have to check it out.
> Thanks Melissa.
>
> Samir, thank you too. I am still waiting for all the data so when I get it
> I will look if there is
> homogeneity of the cases and then run the statistics and everything. I do
> not believe that I will
> have more cases of addicts. 25 if as high as it goes.
>
> Samir
>
> GfK BH
> Centar za istrazivanje trzista i javnog mnijenja d.o.o.
>
> Samir Omerovic
> Istrazivac/Researcher
>
> tel. +387 33 550300
> fax. +387 33 444226
> www.gfk.ba
> www.gfk.com
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
> Melissa Ives
> Sent: Wednesday, May 09, 2007 5:51 PM
> To: [hidden email]
> Subject: Re: Methodology?
>
> You may want to look into using propensity scoring to find a comparable
> group of women from the
> 20000 to use as the comparison group. The idea is to balance groups.
> Briefly, you would identify
> characteristics of the women that would predict their likelihood of being
> in the addicted group of
> 25. Create this 'propensity for being in the addicted group' score for
> the 20000 women, and select
> a group from there that closely matches your 25 addicted women.
>
> For more information about how this works, see:
> Luellen, J.K, Shadish, W.R., Clark, M.H. (2005) Propensity Scores: An
> Introduction and Experimental
> Test. Evaluation Review 29(6): 530-558.
> http://intl-erx.sagepub.com/cgi/content/abstract/29/6/530
>
> Melissa
>
> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
> samir
> Sent: Wednesday, May 09, 2007 2:07 AM
> To: [hidden email]
> Subject: Re: [SPSSX-L] Methodology?
>
> Hi!
>
> I suppose looking at the differences between readings from two groups of
> this kinda disproportionate
> samples is not worthless as long as the two groups concerned are stable
> within the groups. You have
> mentioned that the counter part of the 25 women were addicted to certain
> drugs, so you should look
> into the homogeneity of their cases. If they are, this small group also
> give you consistent readings
> whatever you are interested to look into across the groups. Though 25
> against 20,000 seems to be
> unparallel - we will have to accept the scarcity of the later case ( I
> guess that is the situation -
> not too many pregnant women come with drug addiction!)
>
> What you can do if the question still remains is you can try to increase
> the sample from 25 to 30
> plus. If you can raise this to 50 to 100 - nothing like it.
>
> Regards,
> Samir Kr Paul
>
> ----- Original Message -----
> From: "Samir Omerović" <[hidden email]>
> To: <[hidden email]>
> Sent: 09 May, 2007 12:32 PM
> Subject: Methodology?
>
>
>> Dear all,
>>
>>
>>
>> I have data for approximately 20.000 women that gave birth from 2001
>> to
>> 2007 in one hospital. I also
>> have 25 (only) women who have declared themselves as addicted to
>> certain drugs and who gave birth from 2001 to 2007 in the same
>> hospital (which makes it roughly about
>> 20.025 women in total). I would
>> like to test some differences in the weight of the child, blood cells,
>> etc... between addicts and non-addicts (between 20.000 women and 25
>> women). I am not sure if this is ok thing to do to use t-test, anova,
>> chi square, etc. with such disproportional size samples.
>>
>> Any thoughts on the subject?
>>
>>
>>
>> Samir
>>
>>
>>
>>
>>
>>
>>
>> GfK BH
>> Centar za istrazivanje trzista i javnog mnijenja d.o.o.
>>
>> Samir Omerovic
>> Istrazivac/Researcher
>>
>> 71000 Sarajevo, Skenderija 44
>> BA, Bosna i Hercegovina
>> Tel. +387 33 550300
>> Fax. +387 33 444226
>> www.gfk.ba
>> www.gfk.com
>>
>>
>> --
>> No virus found in this incoming message.
>> Checked by AVG Free Edition.
>> Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date:
>> 05-May-07
>> 10:34 AM
>>
>
>
> PRIVILEGED AND CONFIDENTIAL INFORMATION
> This transmittal and any attachments may contain PRIVILEGED AND
> CONFIDENTIAL information and is intended only for the use of the
> addressee. If you are not the designated recipient, or an employee
> or agent authorized to deliver such transmittals to the designated
> recipient, you are hereby notified that any dissemination,
> copying or publication of this transmittal is strictly prohibited. If
> you have received this transmittal in error, please notify us
> immediately by replying to the sender and delete this copy from your
> system. You may also call us at (309) 827-6026 for assistance.
>
>
> --
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.5.432 / Virus Database: 268.14.12/790 - Release Date: 05-May-07
> 10:34 AM
>

Ornelas, Fermin

Re: Methodology?

After being burned with a normality question I promised myself I would avoid responding to any posting. But here I go again.

In the credit card industry usually the number of bads is relatively small and what one usually does is to select a sample and randomly assign the bads to two sub samples for development and validation. But because the total bads is too small a weight is assigned to each observation. Then the probability of default is estimated using logistic regression. It seems that the problem at hand could be approached in a similar manner. However with 25 cases I am not sure that the results could be robust enough, but it is worth the try.

Fermin Ornelas, Ph.D.
Management Analyst III, AZ DES
Tel: (602) 542-5639
E-mail: [hidden email]

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of samir
Sent: Thursday, May 10, 2007 2:23 AM
To: [hidden email]
Subject: Re: Methodology?

prop scoring sounds to be more methodical, but I guess you can at least
approach in line with Melissa's suggestion. Try to match a similar sample
for the addicted women that you finally come up in terms of demographics -
age, race ( if there is any) and other important socio-economic parameters
like the other person has also suggested - residence area/status, occupation
etc. This wd help increase between groups homogeneity and in case there is a
lack of it, SPSS ANOVA can address the issue with further options available
there.

All the best!

Samir

----- Original Message -----
From: "Samir Omerović" <[hidden email]>
To: <[hidden email]>
Sent: 10 May, 2007 2:35 PM
Subject: Re: Methodology?

NOTICE: This e-mail (and any attachments) may contain PRIVILEGED OR CONFIDENTIAL information and is intended only for the use of the specific individual(s) to whom it is addressed. It may contain information that is privileged and confidential under state and federal law. This information may be used or disclosed only in accordance with law, and you may be subject to penalties under law for improper use or further disclosure of the information in this e-mail and its attachments. If you have received this e-mail in error, please immediately notify the person named above by reply e-mail, and then delete the original e-mail. Thank you.

Max Jasper

Re: Methodology?

In reply to this post by Samir Omerovic

He just meant "pregnant rat women", a bad translation from his mother
tongue. correct would be: "pregnant rat females"!