[Rasch] split half reliability

Gregor Sočan gregor.socan at ff.uni-lj.si
Fri May 22 18:07:40 AEST 2015


Hello,

I do not agree with Tom. It is certainly not a bogus measure, except if 
you consider the classical test theory to be a bogus measurement 
paradigm. ;-)  Split-half reliability basically provides the same kind 
of information as alpha, that is, a lower bound to the sample 
reliability. Of course, both of them make sense only in the classical 
test theory framework.
Parisa: Split-half gives more accurate results than alpha when the test 
items are relatively heterogeneous and you can divide them into two 
similar groups: so, I guess that in your case split-half might really be 
more approapriate than alpha (although I do not know what exactly is 
meant by NRT and CRT - are these measurements comparable to conventional 
test items?).
The original reference is Guttman's (1945) paper in Psychometrika.

Best regards, Gregor


Dne 22.5.2015 ob 8:27 je Tom Conner zapisal(a):
> Parisa,  I don't know the answer to your specific question.  But I do
> know that split half reliability is a bogus measure.  It tells you
> nothing that is not contained in the variance.  I recommend you not use
> it unless a journal requires it.
>
> tlc
>
> On 5/22/15 4:39 AM, Parisa Daftari Fard wrote:
>> Dear Friends,
>>
>> I apologize to interrupt the ongoing discussion. I have been reading articles on dynamic assessment. One facial difference between such papers and others is that they reported split half reliability instead of Alpha. Is this due to the fact that DA is half NRT and Half CRT? I appreciate it if you help me with your incisive suggestion and readings.
>>
>>
>> Cordially yours
>> Parisa
>>
>> Parisa Daftarifard
>> IAU (South Tehran Branch)
>> ________________________________________
>> Rasch mailing list
>> email: Rasch at acer.edu.au
>> web: https://mailinglist.acer.edu.au/mailman/options/rasch/connert%40msu.edu



More information about the Rasch mailing list