[Rasch] Logit scores

Snider-Lotz, Tom TSnider-Lotz at qwiz.com
Wed Dec 21 07:45:26 EST 2005


This problem is one reason that large testing programs such as the SAT and ACT exams employ such exotic score scales (200 to 800, and 1 to 36, respectively).  Within my company we have discussed this issue at great length, with no clear solution.  A "1 to 36" scale solves the problem of people trying to relate their scores to classroom grades, but then they have the opposite problem:  a lack of a familiar framework for interpreting the numbers.  
 
  _____  

THOMAS G. SNIDER-LOTZ, Ph.D.
MANAGER, TEST DEVELOPMENT AND ANALYSIS
QWIZ, INC.
1805 OLD ALABAMA ROAD, SUITE 150
ROSWELL, GA 30076
 
PH. 678.832.0555
FX. 770.642.6115
WWW.QWIZ.COM <http://www.qwiz.com/> 
TSNIDER-LOTZ at QWIZ.COM
 
 
 

-----Original Message-----
From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au]On Behalf Of Barth Riley
Sent: Tuesday, December 20, 2005 3:27 PM
To: rasch at acer.edu.au
Subject: RE: [Rasch] Logit scores



I have encountered this problem myself. While 0-100 is a nice "intuitive" scoring convention, it's resemblance to percentages is problematic. Perhaps another scaling range would avoid this problem. Another approach would be to create criterion-based norms by having clinical experts identify the ranges for low, medium, high satisfaction. Showing a variable map might also help to provide a context for the measurements. The interpretation of change along a measurement needs to take into consideration whether we are talking about individual change or group/sample change. While it may be possible for an individual's measure to increase from say, 50-60 up to an 80-90, mean change of this magnitude is far less likely for a group (e.g., hospital unit). How one conveys this to end users is another question!

 

Barth

 

 

 

 

Barth Riley, Ph.D.

Res. Asst. Professor & Associate Program Director

Dept. of Disability and Human Development M/C 626

University of Illinois-Chicago

1640 W. Roosevelt Rd.

Chicago, IL 60608

Voice (312) 355-4054

Fax:   (312) 355-4058

Email: barthr at uic.edu

 


  _____  


From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au] On Behalf Of Lee, Wynne
Sent: Tuesday, December 20, 2005 1:01 PM
To: Mike Linacre (RMT); Rama Guinda; rasch at acer.edu.au
Subject: RE: [Rasch] Logit scores

 

Hi, all

 

I work at PeaceHealth, a health care organization with 6 hospitals and numerous clinics. We routinely used Rasch-based measures of patient satisfaction for several years now, doing what Mike suggested: transforming logits to "zero - 100" scores.  It's a great, logical, rational and easily implemented approach.

 

However, there's a real challenge with this transformation. Our patient satisfaction scale is very good. Of the 2000-odd patients we survey every 6 months, there are some who report near-perfect scores and some near the bottom, but most most fall somewhere around the middle (as distributions will).  

 

The challenge with the "zero-100" representation is that many units for which we issue reports now get scores in the 50 - 75 range. Most people in the medical world are highly educated in very competitive environments. They are used to getting very high grades ... typically 90%+. So, receiving patient satisfaction scores of 50-70 feels like failing.  PeaceHealth has tried, over the years, many strategies, metaphors, discussions, images etc to move people away from that feeling.  It also does not reward particular score levels, but rewards improvements in scores over time. That is, shifts from 25 to 28, 45 to 48 or 75 to 85 are equally valued & rewarded.  

 

Despite those efforts, managers who get the reports often struggle when they hear their unit's score moved to 62.1, up from 58. It still feels like failure or 'barely passing'. That feeling is augmented by having friends at other hospitals that use non-Rasch based commerical patient satisfaction surveys, who often get nice, high scores like 85 or 90!  Competitive urges (and economics) combined with years of conditioning to think that only scores > 80 are really "good" takes a real psychological toll.

 

Maybe this is just a problem for health care organizations.  Has anyone else run into this challenge? Found a work-around?


Wynne 

Wynne A. Lee, Ph.D. 
Data Analyst, Survey Team 
HID Continuous Improvement Division, PeaceHealth 
Bellingham, WA  98225-1898 
Tel. (360) 738-6300 X4581 
Email:  < mailto:wlee at peacehealth.org> 
There can be no happiness if the things we believe in are different from the things we do.            Freya Stark 

 

 

 

 

  -----Original Message-----
From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au]On Behalf Of Mike Linacre (RMT)
Sent: Monday, December 19, 2005 8:46 AM
To: Rama Guinda; rasch at acer.edu.au
Subject: Re: [Rasch] Logit scores

Dear Rama:

How about linearly transforming logits into the range 0-100 for the lowest to highest possible measure on the test? Then the Rasch measures mean "per cent of the way up the test".

What transformations have other folks found useful?

Mike L.

Rama Guinda wrote:



How to report the logit scores in the form of understandable and meaningful scores for the students, parents and teachers?
Scores that are easy to make sense of and at the same time maintain the interval characteristic of the logit scores.

Mike Linacre
Editor, Rasch Measurement Transactions
rmt at rasch.org www.rasch.org/rmt/ Latest RMT: 19:3 Winter



------------------------------------------------------------___________________________________________________________
This message is intended solely for the use of the individual and entity to whom it is addressed, and may contain information that is privileged, confidential, and exempt from disclosure under applicable state and federal laws. If you are not the addressee, or are not authorized to receive for the intended addressee, you are hereby notified that you may not use, copy, distribute, or disclose to anyone this message or the information contained herein. If you have received this message in error, immediately advise the sender by reply email and destroy this message.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20051220/92b85830/attachment.html 


More information about the Rasch mailing list