[Rasch] Specificity of Facets
chkollias at outlook.com
Thu Mar 2 21:28:53 AEDT 2017
I am pretty confident you can use either Winsteps or Facets for your specified analysis. In Winsteps, add the DIF codes to the person label. In FACETS , there is no need for there to be a rater facet in the model in your case – besides the rater facet is just like any other facet (active or inactive) in an analysis which can be added to the analysis model or left out of it completely. It all comes down to what you wish to investigate. From the sounds of things, you are interested in pairwise interactions (bias) between examinees and your other facets, so run FACETS without adding the rater facet in the model or use Winsteps with DIF codes in person label. Hopefully, your data is connected and you do not have any subset problems and/or you have enough responses for what you are investigating..
From: Rasch [mailto:rasch-bounces at acer.edu.au] On Behalf Of Shirin Shirazi
Sent: 02 March 2017 09:34
To: rasch at acer.edu.au
Subject: Re: [Rasch] Specificity of Facets
Dear Dr Linacre and Dear Edward,
I really appreciate your answers. Perhaps I could not convey what I am getting at.
The question is "can we run facets (for DIF for example) without having any rater-mediated questions?" The test includes MC questions containing either 4, 5 or 6 options.
Sent from my iPhone
On Mar 2, 2017, at 1:05 AM, Edward Li <edwardfli at hotmail.com<mailto:edwardfli at hotmail.com>> wrote:
If you are trying to access the potential bias in questions, a DIF analysis is the first thing came to my mind.
On 2 Mar. 2017 6:01 am, Shirin Shirazi <shirin.shirazi at gmail.com<mailto:shirin.shirazi at gmail.com>> wrote:
Dear Rasch Experts,
The impetus for asking this question is that I have been thinking of the possibility of assessing reading comprehension questions in the format of Multiple choice to find bias of facets (examinees, text type/genre, number of items, field of study, age, length of the passage) through MFRM knowing that the raters do not have a say in MC type in the assessment procedure due to the objectivity of scoring, hence I am wondering another measurement technique should be taken to find it?
As a matter of course, I would hesitate to use either Facets or Winsteps to find the biases induced by the above-mentioned facets?
Your help for my briefing will be highly appreciated. In advance, I express my gratitude for tolerating to read such a weird question.
PhD in TEFL, Assistant Professor
Rasch mailing list
email: Rasch at acer.edu.au<mailto:Rasch at acer.edu.au>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Rasch