COMPARISON OF METHODS TO DETECT ITEM PARAMETER DRIFT IN NABTEB SSCE FOR2012-2015 CHEMISTRY MULTIPLE CHOICE TESTS

Abstract

This study investigated the comparison of methods to detect Item Parameter Drift in National Business and
Technical Examination Board (NABTEB) Senior School Certificate Examinations (SSCE) for 2012 – 2015
Chemistry multiple choice tests. It specifically aimed to ascertain the levels of drifted items and compare the
differences in the number of drifted items across the stated examinations years using four methods (Robust–z, 3-
sigma IRT, Delta plot and Mantel–Haenszel methods) of detecting IPD. This study was guided by four research
questions and one hypothesis. The study adopted the survey research design. The population of the study was
3,239 scores of candidates who sat for the National Business and Technical Examination Board, NABTEB, SSCE
Chemistry multiple choice tests in the first administration and 8,605 Senior Secondary School SS3 Chemistry
students considered for second administration, making a total population of 11,844 in Edo State, Nigeria. The
sample size used for the study was 5,040 which comprised 3,239 and 1,801 candidates’ scores in first and second
administrations respectively. Multistage sampling technique was employed. The instruments used to generate
data were 50-item National Business and Technical Examination Board, NABTEB, SSCE Chemistry multiple
choice test items each for the four years (2012, 2013, 2014 and 2015) making a total of 200 items. Using
Cronbach Alpha, the reliability coefficients of the instruments obtained were 0.83, 0.85, 0.89 and 0.87 for the
respective years. However, the item parameters were estimated from candidates’ responses using eIRT software.
The four methods (Robust- z method, 3 -sigma IRT method, Delta plot method and Mantel -Haenszel methods)
were used to establish the IPD, while descriptive statistics of frequency count and percentage were used to
answer the research questions, and the hypothesis was tested using Chi-square statistics at 0.05 alpha level. The
results that were obtained from the analysis showed that 20 items, 80 items, 65 items and 151 items drifted using
Robust-z method, 3-sigma IRT method, Delta plot method and Mantel-Haenszel method respectively in 2012 to
2015 NABTEB SSCE Chemistry multiple choice tests, and it was also found that there is a significant difference
in the number of drifted items in 2012 to 2015 NABTEB SSCE Chemistry multiple choice test items using Chisquare statistics. Furthermore, it was concluded that Robust-z method was the most stable method because it
flagged the least number of drifted items of all the four methods. Among others, it was recommended that
NABTEB and other examination bodies should use Robust -z method to detect drift items to avoid false
identification of drifted items.