Posted by Austin Fossey
If you are a fanatical follower of our Questionmark blog, then you already know that we have written more than a dozen articles relating to item analysis in a Classical Test Theory framework. So you may ask, “Austin, why does Questionmark write so much about item analysis statistics? Don’t you ever get out?”
Item analysis statistics are some of the easiest-to-use indicators of item quality, and these are tools that any test developer should be using in their work . By helping people understand these tools, we can help them get the most out of our technologies. And yes, I do get out. I went out to get some coffee once last April.
So why are we writing about item analysis statistics again? Since publishing many of the original blog articles about item analysis, Questionmark has built a new version of the Item Analysis Report in Questionmark Analytics, adding filtering capabilities beyond those of the original Question Statistics Report in Enterprise Reporter.
In my upcoming posts, I will revisit the concepts of item difficulty, item-total score correlation, and high-low discrimination in the context of the Item Analysis Report in Analytics. I will also provide an overview of item reliability and how it would be used operationally in test development.