TTest Vs Odds Ratio

[ציפי שוחט‎]

Dear all.

I am involved in a study where the primary variable of interest is Suicide (Yes/No).

The researcher requested a TTest on all of the explanatory variables (Depression Scale etc.) where Suicide is the independant variable. (i.e. the hypothesis tested is that those who commited suicide had higher depression).

In my opinion, I should be calculating Odds Ratios instead of TTests, as Suicide is the dependant variable and the question of interest is whether high depression increases the risk of suicide.

I would be happy to hear you're comments.

Tzippy Shochat

[Ojal john]

Hi Tzippy,

I also think you should be looking at Odds ratios with the other factors (e.g Depression) as independent variables. Looking at levels of say depression by whether one committed suicide or not wont be wrong though, it is informative but avoids the real question which is to quantify the effect of depression (or any other variable) on suicide rate/probability.

Kind regards.

--
--
To post a new thread to MedStats, send email to MedS...@googlegroups.com .
MedStats' home page is http://groups.google.com/group/MedStats .
Rules: http://groups.google.com/group/MedStats/web/medstats-rules
 
---
You received this message because you are subscribed to the Google Groups "MedStats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to medstats+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Peter Flom]

I agree with Tzippy. Not only would logistic regression put the dependent variable where it belongs, but it would allow multiple independent variables to be entered at once (provided the sample size is large enough and the other assumptions are met).

 

Peter

 

Peter Flom

My web site: http://www.statisticalanalysisconsulting.com/

Linked in: http://www.linkedin.com/in/peterflom

Twitter: @PeterFlomStat

--

[John Sorkin]

Tzippy,
A minor suggestion. Be careful! If one does not take great care when writing, it is easy for the reader to confuse Student's t-test  (note the lower case "t") which is a test done on a single dependent variable and a T-test (note upper case "T") which is the multivariate analogue of Student's t-test in which multiple dependent variables are tested at the same time.
John 

John David Sorkin M.D., Ph.D.
Professor of Medicine
Chief, Biostatistics and Informatics
University of Maryland School of Medicine Division of Gerontology and Geriatric Medicine
Baltimore VA Medical Center
10 North Greene Street
GRECC (BT/18/GR)
Baltimore, MD 21201-1524
(Phone) 410-605-7119
(Fax) 410-605-7913 (Please call phone number above prior to faxing)

>>> "Peter Flom" <peterflom...@mindspring.com> 01/09/14 6:39 AM >>>

Confidentiality Statement:

This email message, including any attachments, is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message.

[Peter Flom]

Very astute, John!  I think the vast majority of people mean “t” when they say “t-test” or even when they say “T-test” but it is good to be exact.

 

In addition, some people write computer code in all caps, so that, in SAS for example, they would have PROC T-TEST.

 

Peter

[John Whittington]

At 07:10 09/01/2014 -0500, Peter Flom wrote:
>Very astute, John! I think the vast majority of people mean t when they
>say t-test or even when they say T-test but it is good to be exact. In
>addition, some people write computer code in all caps, so that, in SAS for
>example, they would have PROC T-TEST.

True (albeit without the hyphen) - and if one wanted to add more potential
confusion, one could mention PROC MULTTEST !

Kind Regards,


John

----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: Joh...@mediscience.co.uk
Buckingham MK18 4EL, UK
----------------------------------------------------------------

[John Sorkin]

John,
The hyphen is a matter of usage. One could say Gosset developed a test we call Student's t, or one could say Gosset developed Student's t-test. In the latter case the hyphen indicates that the proper noun Student modifies t-test, not just the letter t.

John

John David Sorkin M.D., Ph.D.
Professor of Medicine
Chief, Biostatistics and Informatics
University of Maryland School of Medicine Division of Gerontology and Geriatric Medicine
Baltimore VA Medical Center
10 North Greene Street
GRECC (BT/18/GR)
Baltimore, MD 21201-1524
(Phone) 410-605-7119
(Fax) 410-605-7913 (Please call phone number above prior to faxing)

>>> John Whittington <Joh...@mediscience.co.uk> 01/09/14 7:23 AM >>>

[John Whittington]

At 07:41 09/01/2014 -0500, John Sorkin wrote:

John, The hyphen is a matter of usage. ...

In general, that's obviously true. However, Peter was citing SAS syntax, and the SAS compiler does not understand "PROC T-TEST", as witness:

8013  PROC T-TEST; run;

ERROR: Procedure T not found.

... but it does understand PROC TTEST.

[roland andersson]

John

May I ask what is a TTest as opposed to a t-test?

Roland Andersson

2014/1/9 John Whittington <Joh...@mediscience.co.uk>:

[John Whittington]

At 15:40 10/01/2014 +0100, roland andersson wrote:
>John, May I ask what is a TTest as opposed to a t-test?

As John Sorkin said, some people use "t-test" to refer to the standard,
univariate "Student's t-test", but use "T-test" to refer to the
multivariate analogue of that test. However,as I said in response to
Peter's comment about SAS, the SAS syntax for a standard t-test uses PROC
TTEST (or proc ttest, or proc Ttest, or whatever - it's not case sensitive).

[ציפי שוחט‎]

Thankyou all for you're helpfull comments.

Tzippy

2014/1/9 Peter Flom <peterflom...@mindspring.com>

[roland andersson]

I saww to late that I referreed to the wrong person. I tried to google
for T-test but did not finf anything different from t.test. Anyway -
if this is a multivariate analogue to t-test I do not understand how
it can be used for testing the distribution of a depression scale
which I doubt is a normal variable.

Roland

2014/1/10 John Whittington <Joh...@mediscience.co.uk>:

[John Whittington]

At 17:08 10/01/2014 +0100, roland andersson wrote:
>I saww to late that I referreed to the wrong person. I tried to google
>for T-test but did not finf anything different from t.test.

Try Googling "Hotelling T-Test", which I think you'll find almost
invariably uses the upper-case "T", or "multivariate T-test".

>Anyway - if this is a multivariate analogue to t-test I do not understand how
>it can be used for testing the distribution of a depression scale
>which I doubt is a normal variable.

I don't think that anyone has suggested that it can be used for that
purpose. John S was merely pointin out the potential confusion that could
result from use of the upeprcase 'T'in the original question.

[John Sorkin]

John

From one John to another, well said!

John

 

John David Sorkin M.D., Ph.D.
Professor of Medicine
Chief, Biostatistics and Informatics
University of Maryland School of Medicine Division of Gerontology and Geriatric Medicine
Baltimore VA Medical Center
10 North Greene Street
GRECC (BT/18/GR)
Baltimore, MD 21201-1524
(Phone) 410-605-7119
(Fax) 410-605-7913 (Please call phone number above prior to faxing)

>>> John Whittington <Joh...@mediscience.co.uk> 1/10/2014 11:18 AM >>>

--
--
To post a new thread to MedStats, send email to MedS...@googlegroups.com .
MedStats' home page is http://groups.google.com/group/MedStats .
Rules: http://groups.google.com/group/MedStats/web/medstats-rules

---
You received this message because you are subscribed to the Google Groups "MedStats" group.
To unsubscribe from this group and stop receiving emails from it, send an email to medstats+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Swank, Paul R]

Given that the sign is somewhat ambiguous for the multivariate T, I was taught that it was Hotelling's T squared.

Paul

Sent from my iPad

> --
> --
> To post a new thread to MedStats, send email to MedS...@googlegroups.com .

> MedStats' home page is https://urldefense.proofpoint.com/v1/url?u=http://groups.google.com/group/MedStats&k=yYSsEqip9%2FcIjLHUhVwIqA%3D%3D%0A&r=MsTc3vpmGdusVsnu9TIMrF5Pxq5xArSzq1fV4Jqks9o%3D%0A&m=peg8uNqfwO3x7%2FyiYTbE0YxKdflAlM44%2F85DVa9fk%2Bo%3D%0A&s=e38a8c843936e6d624b5dadf4f5a22775beb62be3e79b3a0419270c5c513e71f .
> Rules: https://urldefense.proofpoint.com/v1/url?u=http://groups.google.com/group/MedStats/web/medstats-rules&k=yYSsEqip9%2FcIjLHUhVwIqA%3D%3D%0A&r=MsTc3vpmGdusVsnu9TIMrF5Pxq5xArSzq1fV4Jqks9o%3D%0A&m=peg8uNqfwO3x7%2FyiYTbE0YxKdflAlM44%2F85DVa9fk%2Bo%3D%0A&s=bc3c53b47a33aaacf3bc6b248648d32fd26c9feffa1904c6299be3861070c4c3

>
> ---
> You received this message because you are subscribed to the Google Groups "MedStats" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to medstats+u...@googlegroups.com.

> For more options, visit https://urldefense.proofpoint.com/v1/url?u=https://groups.google.com/groups/opt_out&k=yYSsEqip9%2FcIjLHUhVwIqA%3D%3D%0A&r=MsTc3vpmGdusVsnu9TIMrF5Pxq5xArSzq1fV4Jqks9o%3D%0A&m=peg8uNqfwO3x7%2FyiYTbE0YxKdflAlM44%2F85DVa9fk%2Bo%3D%0A&s=f8a1b3e3a0ecc912e6f4be3d669e8d280f69800de293962c8750f568f208b42c.