DOWNLOAD THE COMPLETE TECHNICAL QUALITY DOCUMENT (900K PDF)
		Download this topic only (220K PDF)
	Nationally Tested Research Design The tryout phase for TerraNova, The Second Edition involved large samples of students across the country and provided empirical information about the quality of the tryout items. On the basis of these data, the best items were selected to appear in the final assessments.
		Item Tryout - All of the items developed for TerraNova, The Second Edition were administered to more than 100,000 students in public and private schools across the country in the spring and fall of 1998. Data obtained in the tryout studies were analyzed and the results used to ensure selection of the most accurate, consistent, and equitable items possible.
		Parameter Estimation - Assessments in the TerraNova family include both selected-response and constructed-response items. Because these types of items have different characteristics, two different Item Response Theory (IRT) models were used in analysis of the tryout data. A three-parameter logistic (3PL) model was used for selected-response items. A two-parameter partial credit (2PPC) model was used for constructed-response items.
		Differential Item Functioning Analysis - To minimize item bias, the tryout items were analyzed for differential item functioning (DIF). Item DIF ratings were used in constructing the test to produce final tests with average DIF ratings that were better than the average DIF rating for the entire pool of available items.
		Item Selection - TerraNova, The Second Edition was designed to be an alternate form of the TerraNova family. Developers selected items with statistical characteristics that were most similar to those of the items in the first TerraNova series. To assist this process, the test developers used ITEMSYS, a PC-based software application developed at CTB, to select the final test items for TerraNova, The Second Edition.