A better grid is posted here:

Keep in mind that this grid was designed to change to a median of 65 ... if 
you wanted to convert to some different median, then the grid would change.

http://roberts.ed.psu.edu/users/droberts/mmm3.jpg

Again, as an example. say the median of YOUR class is 50. We take the 
"strip" marked 90 to 10 ... slide it DOWN until the 50 on the strip (your 
median) lines up with the diagonal, which is the fixed median of 65. Then, 
score by score, you adjust your values to the values on the scale that 
would produce a median of 65. Thus, in this case, the "approximate" 
conversions would be:

Old    100       90       80      70    60      50      40      30     20 
   10     0

New    100       93       86      78    72      65      52      39     25 
   12     0

Of course, 100 and 0 NEVER change [I can see the logic in 100 not changing 
... it was 100% at the beginning but, what if we had an upward adjustment 
like in the example above? Why should not the one getting (who is in deep 
trouble anyway) 0% not get a little bonus?]

The more your median is different than 65, the more the values NEAR the 
median will change. The closer your median is to 65, the less ALL the 
values will change.

As an exercise once, since this adjustment does NOT adjust equally across 
the scale, I looked at the:

1. Difference between old and new scores
2. % of the Difference compared to the new
3. Possible remaining score
4. % of the Difference compared to the possible remaining

And there were other ways you might look at the data.

In any case, such adjustments impact on individual examinees in different 
(in some cases, drastically different) ways.

CERTAINLY was MAGIC!

Dennis Roberts, 208 Cedar Bldg., University Park PA 16802
<Emailto: [EMAIL PROTECTED]>
WWW: http://roberts.ed.psu.edu/users/droberts/drober~1.htm
AC 8148632401

.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to