Learn More About GMAT(3)

Why You Can Rely on GMAT® Scores

We deliver the GMAT® exam under standardized conditions around the world with the highest level of security, maintaining a level playing field so that admissions professionals can reliably compare applicants from a wide variety of academic, work, and personal backgrounds. For more than 50 years, the GMAT exam has been repeatedly studied, tested, and modified to ensure that it continues to meet the assessment needs of graduate management programs.
Test scores are not precise measures—even the best possible test can provide only an estimate of an individual's abilities. Because they are estimates, test scores (like other measures) are subject to a certain amount of chance variation that is inherent in the measurement process.

Reliability

Reliability indicates the probability that a test taker would keep the same score if he or she were to take the test more than once. On a scale where perfect reliability is 1.00, the average reliability for GMAT scores is:

• Total score: 0.92
• Verbal score: 0.90
• Quantitative score: 0.89

As these statistics indicate, the reliability of the GMAT score accurately reflecting abilities is very high.
However, repeat test takers are unlikely to get the exact same scores each time. The chance variation between their scores can be estimated statistically, though, and given a value known as the standard error of measurement.
The current standard error of measurement for the GMAT Total score is 29, which means the reported GMAT Total score is within 29 points above or below a score reflecting true performance—a repeat test taker with Total scores within 29 points of each other knows he or she has performed consistently.
The standard error of measurement for the Verbal section is 2.8, and for the Quantitative section it is 3.0.

Validity

Validity is the degree to which GMAT scores predict first-year grades in graduate management programs. Since 1978, GMAC has conducted many validity studies in which GMAT scores, undergraduate grade point averages, and average grades for the first year of graduate school were analyzed.
The most recent studies indicate that the average correlation between GMAT Verbal, Quantitative, and Analytical Writing Assessment (AWA) scores and mid-program graduate management school grades was 0.48 (where 1.0 indicates perfect accuracy of prediction). The correlation between undergraduate grade point average (GPA) and first-year graduate management school grades was 0.28. When GMAT scores were combined with undergraduate GPA, the median correlation was 0.53.
These results indicate that GMAT scores are generally better than undergraduate GPAs for predicting average grades in the first-year of graduate management school. However, the best predictor is combined GMAT scores and undergraduate GPA.
The GMAC Validity Study Service can help you pinpoint the relationship between GMAT scores and performance in your school’s program. This free service analyzes your admissions criteria and lets you know exactly how valid the GMAT exam is for your program.

Fairness

We take the fairness of GMAT questions very seriously. Three procedures are used in the test development process to ensure that questions are not biased against any sub-population group of GMAT test takers.

1. Writers of GMAT questions are well trained to guard against potentially biased test questions by avoiding topics that tend to cause bias. For example, certain topics may involve vocabulary that is more familiar to test takers from one gender, or geographic or ethnic group, thus making test questions relating to the topics potentially easier for these test takers. Writers of GMAT questions are trained to avoid topics and test questions that may favor or disadvantage particular subgroups of test takers.
2. Before new questions are included in the GMAT exam, they are reviewed by independent panels of fairness experts. If fairness reviewers identify potential issues of sensitivity or bias in a question, it is revised or discarded.
3. Every new question is first asked on a non-scoring basis on the GMAT exam, and test-taker response data are analyzed.  Only after passing statistical criteria are GMAT questions used in calculating GMAT scores. One of the statistical analyses performed on all new GMAT questions is called Differential Item Functioning (DIF), which identifies questions that may favor test-takers from one subgroup over another. Questions flagged by DIF methods are further reviewed by test development experts and then revised, discarded, or approved. Revised items are reanalyzed. Only approved questions are subsequently scored on the GMAT exam.

The GMAT exam is developed using best practice methods to ensure fairness for GMAT test taker subgroups.