psychometrics - need some help

[shock-me-sane]

Does anyone have a good place to reference about using psychometrics for multiple choice? A non-psych teacher is attempting to use a test filled with error and appears unwilling to budge unless the wind flies up her skirt at the right angle.

Thanks!

---

Members don't see this ad.

 

[erg923]

I'm not sure i understand your question. Are alluding to examining the distribution of the scores obtained in the exam (ideally a "bell curve" in undergrad classes), and/or individual item analysis? Or are u referring to the reliability of the test....like internal consistency, split-half reliability, things of that nature? Or maybe you are referring to construct and/or convergent validity of the exam? How is this teachers test filled with "error" exactly? What is wrong with it?

 

[shock-me-sane]

Nevermind. I figured out what I needed. I worded the post badly. Essentially an instructor wrote a very ambigious test and said even though 97% of the class got an item wrong that it was still a good question

I was just looking for some data about how to do item analysis of mc tests to try to reason with her. (and the above was just a specific example which spread throughout a ton of the questions). The whole class is in a bit of a tizzy.

I was looking for item discrimination index and such.

thanks anyhow

 

[erg923]

Try reading this: http://www.eric.ed.gov/ERICDocs/data/ericdocs2sql/content_storage_01/0000019b/80/14/a5/81.pdf