hi from Italy,
question: can someone confirm to me that item discriminative efficiency formula use the covariance between item score and whole test score though I use random items divided in 6 categories? If yes, I would like to have the discriminative efficiency calculated category by category, can I have it?
research idea: I'm making some random tests (in italian) based on the Item Responce Teory for help future university students to become familiar with the multiple choice items. Now i'm testing the the parameters of the items. I use 6 categories -theoretically- orthogonal and to use the disciminative efficency calculated whit the whole test score is useless for me, I need it calculated whit the sub-tests: categories by categories. I can use R to make graphics and the LTM library, can I integrate R whit moodle item's analysis??
My test consists of 60 questions, divided into 6 categories, taken from an archive of 1,000 questions theoretically categorized (biology, mathematics, general knowledge, verbal skills, analytical skills, quantitative skills)
that is my site https://elearning.unipd.it/orientamento/course/view.php?id=127
I can't attach the report file...
thanks for listening, sorry for English
italian translate:
domanda: qualcuno può confermarmi che la formula dell'efficienza discriminativa degli item usa la covarianza tra il punteggio dell'item e il punteggio dell'intero test anche se io uso degli item casuali divisi in 6 categorie? Se si, mi piacerebbe avere l'efficienza discriminativa calcolata categoria per categoria, posso averla?
It is all documented here: http://docs.moodle.org/dev/Quiz_item_analysis_calculations
Decriminitive efficiency is an odd one.
Discrimination index is is probably what you are thinking of: covariance between item score and whole test score.
Decriminitive efficiency is the Discrimination index as a fraction of the maximum possible it could be given the marks for that question. Basically, if you have a question with a very high average score (high facility index) it is very hard to get a high Discrimination index.
I had already read the document that you linked and I was sad because that formula is useless for me. There is a way to have the report of all the raw scores of the items or change the formula? I don't need discrimination compared to the whole test score, but compared to the sub-tests.
You can get the raw scores of items from the 'Grades' report. Open the quiz and go to Quiz Administration -> Results -> Grades. You can export the contents of this report as a spread sheet.
I tried what you say! But in this way i will have the score of the specifics items used in the attempts, but I can't distinguish which items was randomly drawn from the archive. The test's items are never the same sixty
You are going to need sql to get that data out of your database. Here is a sql query that gets all data for a specific attempt which might be a start.