Abstract: Differential Item Functioning (DIF) analysis is a key element in evaluating fairness and validity of tests. Gender is cited as a source of construct irrelevant variance. It plays a significant role in determining the volume of students who pass the tertiary level entrance examinations. By so doing, it causes bias and consequently challenges the inference made based on the instrument used for selection of the prospective students. This study aimed at investigating presence of DIF in terms of gender (an issue to do with examination validation) on an entrance examination for Malamulo College of Health Sciences, as triggered by observed low scores of female examinees. The participants (n = 615) were sampled randomly from examinees who were to sit for the 2017-2018 Malawi School Certificate of Education examinations. X-Calibre 4.2 software was used to produce item parameters (a, b and c) which were used in Raju formula. Gender DIF was detected in 77% of the items, biased towards male examinees. On the basis of the findings, it is concluded that the entrance examinations’ test scores were not free of construct-irrelevant variance and the overall fairness of the test was likely compromised at it showed that it was heavily biased towards male examinees. Analysis of DIF items inform use and interpretation of test. This is dependent on item writing, review and the continuous analysis of the items to improve the validity of the instrument(s). By so doing, the examinees are protected and the information that is needed from the examinations is attained.

Alex Ansaye Mwanyongo – Malawi Adventist University, Department of Biomedical Sciences.

Mzimba South District Laboratory, Mzimba, Malawi.

Elias Bonya – Malawi Adventist University, Department of Biomedical Sciences

Exton Siyano – Malawi Adventist University, Department of Biomedical Sciences

Adam M. Nyanda – Malawi Adventist University, Department of Biomedical Sciences

Jonathan Majamanda – Malawi Adventist University, Department of Biomedical Sciences

Linly Linje – Malawi Adventist University, Department of Biomedical Sciences

Victor Peter Nkungula – Malawi Adventist University, Department of Basic Sciences.

Email: mlozenim@mchs.adventist.org/ mmlozen@yahoo.com

Abstract: This observational study on retrospective and prospective data for 35 months examined whether the laboratory department at Mzimba district hospital ensures a total quality management across all its processes is achieved using the sigma-metric. Bias and bias percentage were calculated using true values collected from external quality assurance results and package inserts. Total allowable errors for each parameter were extracted from the Clinical Laboratory Improvement Amendment (CLIA), with the Coefficient variations extracted from past statistical control values of the same lot numbers and external quality assurance results. Six sigma metrics formula was used for metrics calculation of all 23 parameters for this entire medical laboratory department. All 23 EQC parameters had less than 5 sigma metrics at p value of 0.0000. White blood count (WBC), Mean cell haemoglobin concentration (MCHC), and platelet (PLT) had sigma metrics between 4-5. Red blood cells (RBC), hematocrit (HCT), Mean cell volume (MCV), and haemoglobin (HGB) had sigma metrics < 3. Alkaline phosphatase (ALP), total protein (TP) had sigma between 1-3. Albumin (ALB), total bilirubin (BIL T), direct bilirubin (BIL D), alanine aminotransferase (ALT), aspartate transaminases (AST), Gamma Glutamyl Transaminase (GGT), creatinine (CR) and UREA had sigma metrics less than 3. Low and High parameters for both Viral load and SARS-Cov-2 scored less than 1 sigma metrics. On comparisons between years none of the parameters performed satisfactorily. There is a need to meet high class performance on quality control processes across all quality management system essentials.