MCAT 언어추론 점수 : 영어가 모국어가 아닌 학생에서 의과대학 성적 예측이 떨어짐.

MCAT 언어추론 점수 : 영어가 모국어가 아닌 학생에서 의과대학 성적 예측이 떨어짐.

MCAT Verbal Reasoning score: less predictive of medical school performance for English language learners

Babbi Winegarden,1 Dale Glaser,2 Alan Schwartz3 & Carolyn Kelly4

배경

MCAT점수는 의과대학 입학생 선발 기준에 널리 사용된다. 영어를 외국어로 배운 학생은 모국어가 아닌 언어로 시험을 보는 것에 있어서 불이익이 있을 수 있다. English language learners(ELL), 즉 영어를 11세 이후에 배운 학생과 그렇지 않은 학생간 MCAT의 언어추론(verbal reasoning, VR) 점수에서 유의미한 차이를 보인다는 연구결과가 있다. 이번 연구의 목적은 VR 점수와 의과대학에서의 성적간 관계가 ELL과 non-ELL에서 서로 다른 경향을 보이는지를 보고자 하는 것이다.

방법

University of California San Diego School of Medicine admissions files and the Association of American Medical Colleges database로부터 MCAT VR시험과 학생의 의과대학 성적(평점, 시험점수 등)을 1998년~2005년 입학생(2002년~2009년 졸업생) 924명의 학생으로부터 얻었다. 회귀분석 모델을 활용하여 MCAT VR시험점수가 의과대학 성적을 비슷하게 예측하는지를 비교하였다.

결과

ELL여부에 따라 임상실습 전 성적, 학업 우수성, USMLE Stpe 2 CK점수, 두 개의 임상실습 shelf examination 등등을 비롯한 몇 가지 결과에 대해서 VR점수의 예측 정도가 크게 차이가 있었다. ELL학생보다 non-ELL학생에서 VR점수와 의과대학 수행능력간의 높은 상관관계가 나타났다.

결론

MCAT VR점수는 의과대학에서 ELL지원자를 평가할 때 좀 더 신중하게 사용되어야 할 것이다.

---

Data analysis

A hierarchical approach was used for testing a moderated multiple regression model. For each outcome variable, we first entered ELL status (ELL or non-ELL) and MCAT VR scores as predictors, and then, in a second step, entered the interaction between the two predictors into the model. Scores on the VR sub-test were centred in order to minimise non-essential collinearity.10,11

Linear regressions were performed when outcomes were continuous and expected to be normally distributed. When the outcome variable was binary (e.g. pass ⁄ fail), multiple logistic regression was conducted. These analyses were conducted using SPSS Version 18.0.2 (SPSS, Inc., Chicago, IL, USA). When outcomes were counts of unusual events (e.g. number of failed courses), zero-inflated Poisson regression was performed using Mplus Version 6.1 (Muthe´n &Muthe´n, Los Angeles, CA, USA).

In these models, the key test for the differential predictive ability of the VR sub-test on medical school outcomes for ELLs and non-ELLs is whether the interaction term is significant. In the coding of our data, a significant negative interaction would be interpreted as a finding that the relationship between VR score and outcome is stronger for non-ELLs than for ELLs. When interactions were significant, we examined the associations between VR score and outcome within each group.

---

Med Educ. 2012 Sep;46(9):878-86. doi: 10.1111/j.1365-2923.2012.04315.x.

MCAT Verbal Reasoning score: less predictive of medical school performance for English language learners.

Winegarden B, Glaser D, Schwartz A, Kelly C.

Source

Division of Medical Education and Department of Psychiatry, University of California San Diego, La Jolla, California 92093-0092, USA. bwinegarden@ucsd.edu

Abstract

CONTEXT:

Medical College Admission Test (MCAT) scores are widely used as part of the decision-making process for selecting candidates for admission to medical school. Applicants who learned English as a second language may be at a disadvantage when taking tests in their non-nativelanguage. Preliminary research found significant differences between English language learners (ELLs), applicants who learned English after the age of 11 years, and non-ELL examinees on the Verbal Reasoning (VR) sub-test of the MCAT. The purpose of this study was to determine if relationships between VR sub-test scores and measures of medical school performance differed between ELL and non-ELL students.

METHODS:

Scores on the MCAT VR sub-test and student performance outcomes (grades, examination scores, and markers of distinction and difficulty) were extracted from University of California San Diego School of Medicine admissions files and the Association of American MedicalColleges database for 924 students who matriculated in 1998-2005 (graduation years 2002-2009). Regression models were fitted to determine whetherMCAT VR sub-test scores predicted medical school performance similarly for ELLs and non-ELLs.

RESULTS:

For several outcomes, including pre-clerkship grades, academic distinction, US Medical Licensing Examination Step 2 Clinical Knowledge scores and two clerkship shelf examinations, ELL status significantly affects the ability of the VR score to predict performance. Higher correlations between VR score and medical school performance emerged for non-ELL students than for ELL students for each of these outcomes.

CONCLUSIONS:

The MCAT VR score should be used with discretion when assessing ELL applicants for admission to medical school.

© Blackwell Publishing Ltd 2012.

PMID:

 

22891908

 

[PubMed - indexed for MEDLINE]