Two decades of Applications of Item Response Theory Starting up, benefits, expectations, deceptions and developments. Frans Kleintjes, Cito Relevant Numbers • 1953 • 1981 • 1988 • 500 • 20 • 3 Relevant Numbers • 1953 My year of birth • 1981 Masters Application of a LLTM • 1988 Started with Cito- PRC • 500 Number of employees Cito • 20 Number of employees PRC • 3 Man-years research PRC Expectations • 1953 Dr. Frederic Lord: • Ability scores are test independent • Estimated ability scores to be independent of choice of test items • Item and test characteristics to be sample independent Lord, F.M. (1953) The relation of test score to trait underlying the test EPM 13 (My) Applications of IRT • Cito Reading Index Primary Education • Large scale applications School leaving test primary education Entreetoets 7 • International consultancy Cito Reading Ability Item Text 04 Text 20 Text 43 Large scale applications • 1988=> National assessment program PPON five year cycle; 16 topics ; sample based • 1986=> Student Monitoring System primary education per topic two tests each year; pre internet: computer program; used by almost all schools • 1993-2002 Basic Education age 15; 70 tests per year; each test in thee levels; all subjects covered; In 1999 standard setting program for all subjects based on bookmark method using IRT (=OPLM) results Large scale applications • 2002=> Student Monitoring System secondary education: Allocation+ monitoring progress Four tests, each in thee levels; 5 subjects; used by one third of schools; reporting is internet based • 2008=> Test for children with special needs IRT equating and reporting on scale of ‘regular’ primary education:, Improvement of itemconstruction Large scale applications School leaving test primary school (age 12) • Purpose: Advise on track in secondary education, School (self)evaluation; • High stake, 145 000 students (85% of population) • New test each year with 200 items: language, maths, social science; 11 topics. • Equate total test score over years to report ‘standard score’ using pretest-design Large scale applications School leaving test primary school (age 12) • All items are pre-tested twice in incomplete design; 23 booklets 180 items per booklet; (about 2000 in one pretest) from 2003 IRT based • Deception: unable to predict test characteristics for all topics from pretest, we still need the results on the test. Effect of pre-testing for high stake testing • IRT is used to equate related tests: ‘catch up test’ ; easier version; CBT version; embedded anchortest; to relate this test to other tests Entreetoets Groep 7 (at age 11) • Purpose: to provide overview of student and school achievement (profile) • Language; Maths; Social science: 450 items, 13+3 topics, 130000 students (75% of population) • Renewed every 5 years • ‘Embedded’ field test in 2009 (avoiding pretest effects) • 9 versions, each 200 ‘old’ 250 ‘new’ items => optimal design • IRT ‘OPLM hammer’ to equate these versions to the ‘Entreetoets 7’ by topic for reporting • Evaluation in 2010 • Thank you, • Questions ?
© Copyright 2026 Paperzz