| Summary: | This study has focused on the investigation of the reasons for aberrant response patterns in classroom maths tests. Data were collected from high schools in Cyprus over two academic years. The assessment instruments used included: three Maths Tests, a Test Anxiety Inventory (TAI) and a shorter version of it, an Attention Deficit Hyperactivity Disorder (ADHD) scale and a Maths Self-Esteem Scale. Results showed no associations between any of the factors investigated and misfit when tests with polytomous items were used. Factors investigated included: student and teacher gender, item order, different schools, different teachers, ability, test anxiety, ADHD, maths self-esteem, motivation, language competency, interest in maths, private tuition in maths, study time and class revision. This finding has led to the investigation whether misfit is an inherent characteristic of students and the conclusion that it is not. The only factors that showed some association with misfit were ability (p = 0.022), the interaction of gender with test anxiety (p = 0.018) and different teachers (p = 0.027), and the first two were only for the test containing 12 (out of 16) dichotomous items. Further investigation of these factors is suggested. Analyses of interviews of 21 misfitting students showed that the main reason given for unexpected responses among high ability students was, as expressed by them, carelessness and among low scorers prior knowledge and to a lesser degree cheating and special preference. The two mean square statistics, infit and outfit were also investigated, and an explanation is given for why high infit is considered more of a threat to measurement than high outfit. The researcher finally argues that students with misfitting patterns with high outfit values should not be considered as invalidly measured without further investigation. Similarly, items with high outfit should not be considered as malfunctioning and removed without further investigation
|
|---|
