MSSTL10-Carlow IT May 2010 Setting the scene Initial phase of research Aim of presentation Profiling at risk students Predicting failure of at risk students Conclusions Implications for future work Data collected on Technology and Science students since 1997 Up to 600 students tested each year It currently consists of information on almost 7000 students The dataset contains information on students such as: Gender Leaving Cert. mathematics Grade, Level and Points Degree programme of study Attendance at support tutorials Performance in service mathematics examinations Standard or Non standard Performance in the diagnostic test Numbers at risk of failing service mathematics? 1998 2008 Technology % at risk 32.8% 46.4% (% with HL) 41.0% 33.0% Science %at risk 21.3% 46% (% with HL) 55.0% 38.0% Decline in mathematical competencies between 1997-2008 evident (Gill et al., 2010) Investigation in changes in competencies between 1997-2008 by Leaving Certificate grade(Faulkner et al., 2010) 80 70 Mean percentage correct 60 50 HC1 OA1 40 OA2 OB1 30 OB3 20 10 0 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Year Mean diagnostic test score (expressed as a percentage of correct answers out of 40 questions) from 1998 to 2008 for all students with grades HC1, OA1, OA2, OB1 and OB3(Faulkner et al 2010). Whole Cohort Year 1998 2008 % doing HL 237 (46.7%) 239 (35.3%) % doing OL 266 (52.5%) 381 (56.3%) Non Standard students 4 (0.8%) 57 (8.4%) Total 507 (100%) 677 (100%) To use information on students within the database such as gender, Leaving Certificate points, diagnostic test result etc. to build a predictive model of success/failure Science maths 1 Mean CAO maths points 55.3 53.3 54.1 2006 2007 2008 Statistically significant associations were found between success/failure in Science maths 1 and - CAO maths points - Leaving Certificate Level and Grade - Mean Diagnostic Test results - Students who sat the diagnostic test/did not sit the Diagnostic test Technology maths 1 Mean CAO maths points 55.1 55.5 54.2 2006 2007 2008 Statistically significant associations were found between success/failure in Technology maths 1 and: - CAO maths points - Leaving Certificate Level and Grade - Diagnostic Test result - Students who sat the diagnostic test/did not sit the Diagnostic test Discriminant Function Analysis Why use Discriminant Analysis? It is common practice to use discriminant analysis where there are just two populations The discriminant function analysis can act as a tool for classifying future students The nature of discriminant analysis i.e. its ability to determine what variables have a relationship with performance and categorise students accordingly is of great benefit to the design, implementation and evaluation of any educational program/policy (Thomas et al 1996) Dataset 1. The Technological 2006-2008 Z = 0.059(Leaving Cert. Maths Points) + 0.065(Diagnostic Test Result) where C= 4.3 Z ≥4.3 Z ≤ 4.3 classified as a success classified as a failure. Technology Science % of correctly classified success cases 66.2% 60.3% % of correctly classified failure cases 83.2% 78.0% Overall 69.7% 62.9% Function’s ability to predict failure in 2009 Science and Technology cohorts? Technological Discriminant Function will be used to identify the at risk students entering UL in the academic year 2010/11 Profiling at risk students between 2006-2008 Ordinary Level Leaving Certificate mathematics grade Identified as at risk by the diagnostic test or Have not sat the diagnostic test Predicting failure in service mathematics Discriminant Analysis Technology 2006-2008 function: Z = 0.059(Leaving Cert. Maths Points) + 0.065(Diagnostic Test Result) where C= 4.3 The discriminant function produced in this phase of the research will allow for The identification of at risk students in the academic year 2010/11 in the first week of term The design of a targeted intervention strategy for the identified at risk students