Session Information
Contribution
When examinees take a test their responses are expected to conform to some standard of reasonableness (Smith, 1986): for instance, a response pattern that involves a significant number of wrong answers to `easy' questions but right answers to `hard' questions would be regarded as `aberrant'. Detecting examinees with aberrant response patterns is important as according to the literature their test scores may fail to provide a useful and valid measure of their ability.This study builds on previous work (Petridou & Williams, 2006a; under review) where we have examined the effect of various person- (i.e. ability, gender, language, anxiety and motivation) and class-level factors (i.e. all the above mentioned variables used as class-level variables) on person aberrance with real data under the framework of a two-level (person and classroom) model. This study led to the following main findings: (i) the class contribution to aberrance was found to be quite high; and (ii) there was a strong indication that ability had a statistically significant contribution to aberrance; more able pupils were more likely to produce aberrant response patterns. This study aims to explore further all the aforementioned findings, especially the class-contribution to aberrance by replicating multilevel analyses for test data coming from pupils of various ages from 5 to 14. Test data comes from the Mathematics assessment for Learning and Teaching (MaLT) database. For the purposes of our previous study only a subset of this database was used i.e. the Year 5 group (1300 10-year olds). In this study multilevel analyses will take place with test data coming from all the remaining year groups in the MaLT database i.e. a sample of 11291 pupils aged 5-14. We aim with this replication study to be able to provide answers to the following questionsQ1: Does the classroom have a similar, significant contribution to all age groups?Q2: Is this class contribution stronger in the orally presented tests (pointing to administration and mode of assessment as a key source of variation)? Is there a discontinuity in the pattern of misfit between the (orally presented) tests i.e. MaLT 5-7 (for 5 to 7 year olds) and MaLT8-9 (for 8 to 9 year olds)? Q3: Does the aberrance-ability relation (and the effects of other background variables) change with age? Q4: When joint analyses across year groups are conducted to replicate aberrance findings do these replicate analyses of single separate year groups? (i.e. is the ability finding a test effect?)The methodology to be followed in this study is similar to that in our previous study (Petridou & Williams, under review.). Specifically, two general-purpose fit statistics are used for the identification of aberrant examinees i.e. Infit and Outfit Mean Square. The Infit and Outfit values then become the response variables in two-level models. The only explanatory variables however available to be entered in the two-level models across all year groups are gender and ability as measured from the Rasch model. In this study both joint analyses across year groups and separate analyses for each year group will be employed. This will enable us to examine any differences in misfit findings but more importantly to test whether the significant 'ability' effect found in our previous study is a research artifact. The conclusion will be the relationship between person 'aberrance', ability and age. In particular, this replication study will enable us to examine whether previous findings were sample- and instrument- specific or generalize across other datasets. Petridou & Williams (2006). Identifying, modelling and explaining person misfit. Unpublished. Ph.D. dissertation. University of Manchester. Petridou & Williams (under review). Accounting for real person misfit using multilevel models. (Revise and resubmit for the Journal of Educational Measurement, June 2006). Smith, R. M. (1986). Person Fit in the Rasch Model. Educational and Psychological Measurement, 46, 359-372.Applied Psychological Measurement, Journal of Educational Measurement or Applied Measurement in Education
Search the ECER Programme
- Search for keywords and phrases in "Text Search"
- Restrict in which part of the abstracts to search in "Where to search"
- Search for authors and in the respective field.
- For planning your conference attendance you may want to use the conference app, which will be issued some weeks before the conference
- If you are a session chair, best look up your chairing duties in the conference system (Conftool) or the app.