[Rasch] Item and Person Measures in survey questionnaire

Schulz, Matt mschulz at pacificmetrics.com
Tue Apr 19 23:18:11 EST 2011


Yes, and Yes.

Here is a reference to a paper you might find useful.  I involves a survey of satisfaction with college services.  If I can find an electronic copy, I’ll send it to you directly.

Schulz, E. M., & Sun, A. (2001).  Controlling for rater effects when comparing survey items with incomplete Likert data.  Journal of Applied Measurement, 2 (4), 337-355.

Due to the missing data problem in this study, it was important to distinguish “satisfaction” from “a  propensity to be satisfied”.  But if you aren’t using the Rasch model as a formal solution to a missing data problem, you probably don’t need to be so careful with your terms.  If you have access to a college library or similar resource, here is a link to a form you can use to ask them to subscribe to the Journal of Applied Measurement:

http://www.jampress.org/libform.pdf


From: Juanito Talili [mailto:talilij at yahoo.com]
Sent: Monday, April 18, 2011 11:05 PM
To: Schulz, Matt
Cc: Rasch Forum
Subject: RE: [Rasch] Item and Person Measures in survey questionnaire

May I clarify?    Consider a 15-item instrument that measures satisfaction of food services in a canteen(using 4-point scale from less satisfied to more satisfied). Please correct me if I am wrong.

1. For a person measure (logit), does it mean, on average, the higher the person measure the more satisfied the customers and the lower the person measure the less satisfied the customers?
2. For the item measure (logit), does it mean, on average, the higher the item measure the more difficult the item to endorse by the customers and the lower the item measure the more easier the item to endorse.

Thank you for any comments.

J

--- On Mon, 4/18/11, Schulz, Matt <mschulz at pacificmetrics.com> wrote:

From: Schulz, Matt <mschulz at pacificmetrics.com>
Subject: RE: [Rasch] Item and Person Measures in survey questionnaire
To: "Juanito Talili" <talilij at yahoo.com>, "rasch at acer.edu.au" <rasch at acer.edu.au>
Date: Monday, 18 April, 2011, 2:05 PM

Juanito,



Good question!  It depends on what you are measuring, plus there is a very subtle point to keep in mind.  If your rating scale is something like “How satisfied are you with….”.  You are *not* measuring “satisfaction,” per se but rather, persons’ propensity to be satisfied, or “satisfiability” with the  particular  class of items you are having them rate.   The response to each item tells you how satisfied the person was with that item—and that is data representing a person’s satisfaction.  But the data is not the measure.



This is not so hard to understand when applied to mathematics performance. You can give a person a easy test or a hard test.  A person’s success rate, or ‘performance’ (the data) will be higher on the easy test than on the hard one.  But their mathematics “ability” is the same on both.   It’s the same with a satisfaction questionnaire.  You can apply the questionnaire to items that are in general not satisfying to people (vegetables), or to items that are (candy).   A kid will be very satisfied with candy, but not vegetables.  But their underlying propensity to be satisfied is the same.  You might qualify this trait as propensity to be satisfied with food.  Or even more specifically, “with candy” or “with vegetables”  if you think the trait is not unidimensional across such different classes of food.



In a questionnaire I once worked on, the items were colleges services (e.g., counseling, health care, etc.) and we called the trait, “propensity to be satisfied with college services!”   I suppose, loosely speaking, we could have said we were measuring “satisfaction.”  But it’s important to understand the difference between the data and the measure.



Good luck,



Matt Schulz





From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au] On Behalf Of Juanito Talili
Sent: Monday, April 18, 2011 12:33 AM
To: rasch at acer.edu.au
Subject: [Rasch] Item and Person Measures in survey questionnaire



Dear co-members,



In a right/wrong response format, usually achievement test, the item measures represent the item difficulty while the person measures represent the person ability.



What would be the interpretation of the item and person measures for a case of a survey questionnaire that uses a four-point ordinal scale response format ( strongly disagree to strongly agree). I undertand that the measures are called logits (am I correct?) but I dont know how the item and person measures are interpreted.



Thank you in advance.



J





-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailinglist.acer.edu.au/pipermail/rasch/attachments/20110419/72d285fc/attachment.html 


More information about the Rasch mailing list