Date:         Tue, 5 Feb 2002 13:35:40 -0600
Reply-To:     "R.C. Intrieri" <RC-Intrieri@wiu.edu>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "R.C. Intrieri" <RC-Intrieri@wiu.edu>
Subject:      A statistical Question
Content-Type: text/plain; charset="iso-8859-1"; format=flowed

Dear Colleagues,
This is more of a statistical question rather than strictly an SPSS question.
I am attempting to explain the formula to calculate the Median from Grouped 
data.
The formula for this is the following

Md = Lm + w [(n/2 – fcum)/fm]
Lm = lower limit of the interval that contains the median.
W = width of the interval
fm = is the frequency
fcum = the number of observations falling below this interval.

The formula comes from Glass & Hopkins (3rd edition) Statistical methods in 
education and psychology
EQ 4.1 P.52

The question that I have (and I need a reference if possible) is what is 
the justification from a theoretical point of view for using "the number of 
observations falling below the interval" as opposed to using the number of 
observations falling above the interval?  Any help would be greatly 
appreciated.  Thanks.

Bob Intrieri
Robert C. Intrieri, PhD         Office (309) 298-1336
Department of Psychology        Fax    (309) 298-2179
Western Illinois University     E-mail RC-Intrieri@wiu.edu
1 University Circle
Macomb IL 61455-1390