StuIQ is another real-life example, based on the development of a scholastic aptitude test for use with senior high school students.
Two forms of the test were developed, each with 70 items, each having a mixture of multiple-choice and constructed-response items.
Questions investigated: the reliability of each form as a total test, the reliability of each form's multiple-choice items as subtests, the reliability of the constructed-response subtests, parallel-forms reliability, and practice effects.
A description of the study, "PFExample1", is available in two formats, a webpage and a PDF file. It's a relatively complex study as the item responses were coded in a unique manner. Note that the Excel 2003 version of Lertap was used in this study.
A link to the complete dataset is available on the following page.