NAEP Technical DocumentationIntelligent Character Recognition

The Intelligent Character Recognition (ICR) engine is used to read various hand and machine printing on the assessment questionnaires and supervisor documents. Some information from administration schedules, rosters of questionnaires, and some questions in the school questionnaires are read by the ICR engine and verified by an online key entry operator. Typically, the ICR engine reads over 2 million characters for an assessment cycle. Use of the ICR engine saves NAEP field staff a significant amount of time since they do not have to grid rows and columns of data.

The scanners and programs are also able to read imprinted codes, known as 2-out-of-5 codes, printed on the administration schedule. These 2-out-of-5 codes are imprinted at the same time the booklet ID numbers are printed on the administration schedule, and identify which booklet IDs are listed on the document. (For the 2001 assessment, the same 2-out-of-5 code also identified the demographic information preprinted on the administration schedule when that information was available at the time of printing.) Use of preprinted booklet IDs, demographic information, and 2-out-of-5 codes eliminates a significant amount of online editing time needed to process the NAEP assessments. When the scanning programs are unable to translate the 2-out-of-5 codes, image clips of the booklet ID numbers are displayed to online editing staff for verification.

Last updated 14 January 2010 (GF)

Printer-friendly Version