Demystifying the GMAT: Guarding Against Bias

When I first took the GMAT exam in 1989, almost all test takers were American or English-speaking Canadian citizens seeking to attend school in North America. Only a small number of non-English-speaking test takers who also wanted to attend a North American school took the GMAT exam in those days. Back then, GMAC did not need to be concerned about native English speakers having an unfair advantage. Today, however, slightly more than half of GMAT test takers are non-US citizens, and we at GMAC recognize the importance of providing a test that is fair, valid and reliable for all test takers, regardless of their native language

A fair and reliable test means that all individuals of the same ability should get the same scores. It does not suggest that the averages by native language should be approximately equal. Different groups are expected to have different sets of skills, and within any group, there is also a self-selection bias. Sorry, Argentina, your extremely high mean GMAT Total score does not mean that you are smarter than the rest of us or that the GMAT exam is biased in your favor. It may just mean that in Argentina, only those very qualified for management study actually take the test.

Yes, the GMAT test is administered in English and is designed for programs that teach in English. But the required English skill level is much less than what students will need in the classroom. The exam requires just enough English to allow us to adequately and comprehensively assess Verbal reasoning, Quantitative reasoning and Integrated Reasoning skills. Although the GMAT exam is not a grammar test, it is worth noting that grammar tests do not necessarily favor native speakers, because non-native speakers often learn grammar better than native speakers who go by the “sounds right” rule. I was once turned down for a job as an English language instructor in Italy because my spoken English contained grammar mistakes.

While the GMAT exam tests reasoning and not English, there are several ways bias in favor of native speakers can be introduced into an exam.

1. Use vocabulary, idioms, and constructions that are not universal. My favorite example is skimmed milk. You might expect all educated individuals applying to a graduate level English language program would know skimmed milk. But in India, toned milk and double toned milk are common, and skimmed milk is rare. Vacation and holiday, and quite and very, are other examples. Also under this category are the use of double negatives (The GMAT is not unfair), contractions (should’ve), abbreviations (for example, e.g.), imperatives, possessives, and some sentence structures (Do you have a pencil? Have you a pencil?)

2. Use culturally loaded phrases, idioms, and constructions. Most people who grew up in North America would probably know what is meant by a Mickey Mouse job (a job that is trivial and a waste of time), the whole nine yards (everything), or skedaddle (depart quickly). Also under this category are military terms (your mission), sports terms (game plan), colloquialisms (lots more), regional terms (pop, hoagie), metaphors (launch an idea), euphemisms (under the weather), clichés (between a rock and a hard place), literary references (wears his heart on his sleeve), scriptural references (last straw), US cultural references (Big Mac), lesser known US places and people (McKinley), acronyms (ASAP), and initials (IRS).

3. Use culturally offensive phrases, idioms, and constructions. This includes the use of ethnic slurs, derogatory terms, positive and negative stereotypes, and references that rely on or ignore belief systems (drinking alcohol, eating pork).

Culturally based, loaded, and offensive terms do not always result in differences in performance, but we do not want test takers to be distracted by culturally insensitive questions. We carefully review our questions using criteria defining good item construction. We also compute statistics to assess whether our questions are appropriate across culture groups. We constantly update guidelines for our item writers, including a master list of terms and phrases to avoid in order to assure cultural fairness.

By using carefully defined and thorough item development and review processes, along with statistical analyses to flag questions with possible cultural bias, we have developed a test that minimizes the impact of culture and language. The GMAT exam is the best objective measure of the likelihood of success in management programs across the globe.

If you have any doubts about the validity of the GMAT exam for your program, we urge you to participate in our free GMAT Validity Study Service. In addition to providing valuable statistics about the exam and subgroup performance in your program, this service designed by admissions personnel provides you with the statistically optimal weights for Quantitative, Verbal, Total, undergraduate GPA, and other criteria of your choice, with Integrated Reasoning soon to come. Coupling this information with your assessment of candidate tenacity can help you meet and exceed your admission goals.

Lawrence M. Rudner, PhD, MBA, is vice president of research and development and chief psychometrician for the Graduate Management Admission Council.