OSCATS Basics [Was Issue 3 in oscats: The specified procedure could not be found(libintl_setlocale)]
Michael Culbertson
The issue tracker is for discussing programming errors in OSCATS,
not for how to use OSCATS, so I'm moving this discussion to the
mailing list.
> 1)i want to find Theta(ability) after each item is administered by examinee
> and based on that theta i want to display next item from the bank to
> examinee.
Yes, that's what a CAT does. To understand better how OSCATS
accomplishes this, I suggest reading the documentation for the
OscatsTest class. There is also a conference presentation that
describes some of how OSCATS is put together, available at:
http://hdl.handle.net/2142/27706
> 2)want to find Final Theta after exam is finished.
You can access the final estimated ability with oscats_examinee_get_est_theta().
> 3)want to find index of item in the bank which is displayed to the examinee.
OSCATS currently has built in support only for simulated CAT
administration. If you would like to use it to display items to a
live examinee, you will have to write your own OscatsAlgorithm class
that connects to the OscatsTest::administer signal. See the
presentation above or example 4 (which isn't terribly well commented
yet) for an example of how to create a new algorithm in python.
> can you plz tell me which formula u r using to find Theta(ability) and
> probability of theta.
The formulas for the item response models are listed in the
documentation. Estimation is either by maximum likelihood (using
Newton-Raphson) or Expected A Posteriori (with normal prior), as
selected by the OscatsEstimate:posterior property.
> i am not able to understand the use of some classes like
> 1)covariates
This is for use with explanatory item response models (see Paul de
Boeck, Mark Wilson, Explanatory Item Response Models: A Generalized
Linear and Nonlinear Approach, Springer 2004).
> 2)AlgExposureCounter
This algorithm keeps track of how many times each item is administered
(to any examinee).
> and one more doubt is what is the difference between Simulated theta and
> Estimated theta and why u r finding simulated Theta?
One doesn't "find" simulate Theta, per se---one gives it. OSCATS is
designed to facilitate CAT simulation research, as well as
(eventually) providing a back-end for actual CAT administration. In
CAT simulation, one gives a "true" (or simulated) ability level that
is used to generate the item responses and allows the CAT to operate
based on its own estimated ability. Typically, one is interested in
the difference between the recovered (estimated) ability and the
simulated ("true") ability or the usage of the item bank (hence
AlgExposureCounter).
Michael Culbertson
QUERIES Division
Department of Educational Psychology
University of Illinois, Urbana-Champaign
On Wed, Nov 23, 2011 at 12:09 AM, <osc...@googlecode.com> wrote:
>
> Comment #8 on issue 3 by murukutl...@gmail.com: The specified procedure
> could not be found(libintl_setlocale)
> http://code.google.com/p/oscats/issues/detail?id=3
>
> Hello Michael,
> i want to conduct test using IRT 1 parameter model.for this i
> have an Item Bank with difficulty parameter ranges from(-2 to 2).what my
> problem is
>
> 1)i want to find Theta(ability) after each item is administered by examinee
> and based on that theta i want to display next item from the bank to
> examinee.
> 2)want to find Final Theta after exam is finished.
> 3)want to find index of item in the bank which is displayed to the examinee.
>
> can you plz tell me which formula u r using to find Theta(ability) and
> probability of theta.
>
> i am not able to understand the use of some classes like
> 1)covariates
> 2)AlgExposureCounter
> and one more doubt is what is the difference between Simulated theta and
> Estimated theta and why u r finding simulated Theta?
>
> please suggest me some procedure to conduct the Test
> Thanks in advance
chinni
Dear Michael,
when I call test[i].administer() method it is
internally calling the algorithms and setting Response value ,Theta
values.I am not able to understand how the methods and algorithms are
executed and is there any way to set the Response value (1/ 0)
manually?
In previous post i asked which formula you are using to estimate the
Theta.i read the document but no where i found Theta estimation
formula.
Thanks & Regards
Swathi
chinni
Michael Culbertson
In OSCATS, each algorithm is an object. OSCATS makes use of signals
to indicate when the algorithm should perform its piece of the CAT, as
described in the conference presentation I referenced in my last
message. The order in which the signals are emitted by the OscatsTest
object is listed in the documentation for the oscats_test_administer()
function.
By default, OSCATS works by simulating examinee responses. If you
want to specify the examinee responses manually, you'll need to write
your own administration class (that is, write a class that attaches
the the OscatsTest::administer signal). See example 4
(CustomSimulateAlg) and the conference presentation for an example of
how to do this in python. The basic steps are similar in other
languages.
Ability estimation is by Maximum Likelihood (unless you set
OscatsAlgEstimate:posterior to true, in which case it is by expected a
posteriori with normal prior):
http://en.wikipedia.org/wiki/Maximum_likelihood
The likelihood is maximized by the Newton-Raphson algorithm:
http://en.wikipedia.org/wiki/Newton-Raphson
Michael Culbertson
QUERIES Division
Department of Educational Psychology
University of Illinois, Urbana-Champaign
> --
> You received this message because you are subscribed to the Google Groups "OSCATS Discussion Group" group.
> To post to this group, send email to osc...@googlegroups.com.
> To unsubscribe from this group, send email to oscats+un...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/oscats?hl=en.
>
>
chinni
Dear Michael,
I am developing my own CustomSimAlgorithm to set
Response manually by following python example.But how to connect this
algorithm to administer signal.is there any method to connect?
Thanks & Regards
Swathi