median of two groups are statistically significant

[Nabaneeta Saha]

Hello All,

What is the right statistical test to determine whether the median of multiple
groups are statistically significant or not?

Thank you for your help,

Nsaha6

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

What is the nature of the data?  Is it reasonable to assume that the distribution shape is the same in both (or all) groups?  To see why I ask that question, see the "Do non-parametric tests compare medians?" section in Chapter 10 of "Statistics at Square One".

http://resources.bmj.com/bmj/readers/readers/statistics-at-square-one/chapter-10-rank-score-tests

HTH.

Nabaneeta Saha wrote

Hello All,

What is the right statistical test to determine whether the median of multiple
groups are statistically significant or not?

Thank you for your help,

Nsaha6

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Rich Ulrich]

The chapter cited by Bruce points out that the usual non-parametric tests
are tests on average rank, and not tests on the median.  And those tests
do assume that the original distributions are of the same form, so that
the tests are tests on "shift" or location in similar distributions.

For fewest assumptions, you might test the equality of medians
by a simple contingency table that reports how many in each
group are below/above  the overall median.  (Recode every score
to 0/1 and cross-tabulate.)  This principle can be applied to other
fractions just as readily, such as, "top 1%".

People are usually more interested in the average rank.  There might be
more reports published where people mis-report a rank-test as a test
on medians than where people consciously intend a test on medians
and use an intentional test on medians. 

--
Rich Ulrich

> Date: Mon, 22 Aug 2011 10:44:00 -0400
> From: [hidden email]
> Subject: median of two groups are statistically significant
> To: [hidden email]
>
> Hello All,
>
> What is the right statistical test to determine whether the median of multiple
> groups are statistically significant or not?

[Bruce Weaver]

I just noticed that I received this off-list reply at the hotmail address I use for posting here.  (I only check that inbox occasionally, as it says in my sig file below).  I'm posting it here because others might have some ideas about what can be done.

--- start of off-list reply ---
Hi Bruce,

Thanks! The data does not follow a normal distribution in any of the groups. I am dealing with bacterial HPC counts. It ranges from 0 to 10^10 in different groups. So, I want to compare if the median count in different groups statistically significant or not.

Thanks again

Nsaha6
--- end of off-list reply ---

If I follow, you have k groups, and for each group there is one data-point, which is a bacteria count ranging between 0 and 10^10.  Correct?  This is outside my research area.  What is typically done in your field?  

Bruce Weaver wrote

What is the nature of the data?  Is it reasonable to assume that the distribution shape is the same in both (or all) groups?  To see why I ask that question, see the "Do non-parametric tests compare medians?" section in Chapter 10 of "Statistics at Square One".

http://resources.bmj.com/bmj/readers/readers/statistics-at-square-one/chapter-10-rank-score-tests

HTH.

Nabaneeta Saha wrote

Hello All,

What is the right statistical test to determine whether the median of multiple
groups are statistically significant or not?

Thank you for your help,

Nsaha6

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Greg Hurtz, Ph.D.]

Not sure if this is a resolved thread or not, but the following
reference and link to an Excel file might be helpful:

Bonett, D.G. and Price, R.M. 2002, Statistical Inference for a Linear
Function of Medians: Confidence Intervals, Hypothesis Testing and
Sample Size Requirements.  Psychological Methods, Vol. 7, No. 3,
370-383.

www.gmw.rug.nl/~huisman/spssmanual/medians.xls



***********************************************************************************
Greg Hurtz, Ph.D.
Associate Professor of Psychology, CSUS
Research Specialties:
Industrial Psychology, Statistical Methods, Measurement/Psychometrics
[hidden email] (Consulting/Research)
[hidden email] (Teaching/University Business)
***********************************************************************************




On Mon, Aug 22, 2011 at 7:44 AM, Nabaneeta Saha <[hidden email]> wrote:

> Hello All,
>
> What is the right statistical test to determine whether the median of multiple
> groups are statistically significant or not?
>
> Thank you for your help,
>
> Nsaha6
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Susan Maree Cotton]

 I am currently in Canberra at NHMRC meetings and will have limited access to emails during this time.