[Rasch] Re: CAT ability estimation
garybuck at umich.edu
Wed Jun 4 04:22:47 EST 2008
Thank you for your prompt reply. Given your recommendation to use MLE, I
have a follow up question that is really rather stupidrather like asking
how long is a piece of string, but the issue is important so any response
would be much appreciated.
Since this CAT will be administered over the web, and hence all the
calculations take place in real time on one server, simultaneously for a
number of test takers, do you think that MLE is still feasible? In other
words, are typical modern computers (i.e. A smallish dedicated Sun
web-server) capable of carrying out a number of simultaneous MLE
On 6/2/08 5:29 PM, "Mike Linacre (RMT)" <rmt at rasch.org> wrote:
> Thank you for your email, Gary.
> Ability estimation with known item difficulties:
> MLE method: www.rasch.org/ <http://www.rasch.org/rmt/rmt102t.htm>
> rmt/rmt102t.htm <http://www.rasch.org/rmt/rmt102t.htm>
> Log-odds method: www.rasch.org/rmt/rmt22g.htm
> With modern computers, the more computationally-intensive MLE method is
> practical in real-time applications.
> Mike L.
> At 6/2/2008, you wrote:
>> I am in the process of designing a web-based CAT. I feel I have a fairly good
>> idea of most of the issues involved, except how to estimate the persons Rasch
>> measure based in their responses to a string of calibrated items.
> Mike Linacre
> Editor, Rasch Measurement Transactions
> rmt at rasch.org www.rasch.org/rmt/ <http://www.rasch.org/rmt/> Latest RMT: 21:4
> Spring 2008
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Rasch