[Rasch] Hierarchical Rasch Modeling
ici_kalt at yahoo.com
Sun Jun 9 02:31:21 EST 2013
I have been faced to this kind of comments in a review when doing a SEM and HLM analysis of the data coming from PISA 2009. The review was based on the publications by ACER (main responsible of the PISA project), where they use Conquest and Mplus. Instead we used Winsteps, AMOS and HLM 5.0.
We have discussed to the publishers that it is not necessary to do the analysis exactly as ACER did, using those programs if our model was correct and our approach was well based.
The need of plausible values for PISA is related to the fact that they are doing a sample for every nation and they are not interested on every individual person. We have worked with the plausible values and also other values provided by the PISA data set, but this is useful if the purpose of your project is similar to PISA, otherwise I think this is not necessary.
INSTITUTO DE EVALUACION E INGENIERIA AVANZADA.
Ave. Cordillera Occidental No. 635
Colonia Lomas 4ª San Luis Potosí, San Luis Potosí.
C.P. 78216 MEXICO
(52) (444) 8 25 50 76 / (52) (444) 8 25 50 77 / (52) (444) 8 25 50 78
Página Web (en español): http://www.ieia.com.mx/
Web page (in English): http://www.ieesa-kalt.com/English/Frames_sp_pro.html
From: Wenchung <psywcw at gmail.com>
To: "rasch at acer.edu.au" <rasch at acer.edu.au>
Cc: "<rasch at acer.edu.au>" <rasch at acer.edu.au>
Sent: Friday, June 7, 2013 7:29 PM
Subject: Re: [Rasch] Hierarchical Rasch Modeling
I think the reviewer may be concerned of measurement error in the person measures that are produced in 1st step. When these person measures are directly plugged into the next stage analysis, their measurement errors are ignored, which will underestimate the standard error of the next stage estimates. People often use "plausible values" to resolve with this problem. Please refer to the Pisa technical report or the relevant literature of plausible values.
As I know, mplus or winbugs can do multilevel Rasch analysis.
從我的 iPad 傳送
Robert Hess <Robert.Hess at ASU.edu> 於 2013年6月8日 7:35 寫道：
> First, I disagree with your reviewer. Mike is absolutely correct, if you do
> everything at one time all you do is compound any felonies that might've
> been committed in level I to the point that when you reach level III either
> nothing works or everything works because you missed any possible sources of
> unexplained error by attempting to do everything in a single pass. And you
> are correct, HLM will not do what you are doing nor will it do what the
> reviewer seems to be requesting.
> My best suggestion for you, is to layout the argument detailing clearly why
> you chose to do the three-step process (as you did in your earlier note) and
> find another way to express the sentiment laid out by Mike. Sometimes you're
> dealing with a reviewer who really doesn't understand the model you are
> employing and you have to find a logical explanation why you are sticking to
> your original model - you address this to the editor, but as an explanation,
> not as a failure on the part of the reviewer to understand your model (which
> I will bet dollars to dimes) is what happened.
> Good luck.
> Bob Hess
> -----Original Message-----
> From: rasch-bounces at acer.edu.au [mailto:rasch-bounces at acer.edu.au] On Behalf
> Of Mike Linacre
> Sent: Friday, June 07, 2013 3:56 PM
> To: rasch at acer.edu.au
> Subject: Re: [Rasch] Hierarchical Rasch Modeling
> Stuart and folks:
> The reviewer seems to be addressing the wrong issue: "it would be more
> efficient ..."
> Surely concern should focus on "it would be more accurate ..."
> In science in general, the more accurate approach is "one thing at a time".
> This approach is your 3-level approach where findings at each level can be
> verified before the next level is attempted. In the combined approach,
> inaccuracies at the first level are likely to be invisible by the third
> level producing undetectable distortions in the findings.
> Mike L.
> Rasch mailing list
> Rasch at acer.edu.au
> Rasch mailing list
> Rasch at acer.edu.au
> Unsubscribe: https://mailinglist.acer.edu.au/mailman/options/rasch/psywcw%40gmail.com
Rasch mailing list
Rasch at acer.edu.au
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Rasch