Most of us have seen the “bell curve” before. When data with a “normal distribution” is displayed visually, it looks like an upside down bell, hence the name. “Normal” in this statistical sense means that the data points are not clustered to the left, to the right, or in a random assortment, but in fact have an even appearance that is fat in the middle and thin at both ends. Anytime we collect information and find that it has this normal distribution, we know that it will present a bell curve. Here we will look at the Miller Analogies Test bell curve.
What the bell curve shows
The above picture is not specific or unique to the MAT; it is the idealized shape that most bell curves take. MAT scores happen to fall along a bell curve like the one above. When test scores have a bell curve, we know certain valuable things about them. We know that nearly everyone who takes the exam will score within 1 standard deviation of the median. We can also predict the frequency of scores 2, 3, and 4 standard deviations above that median. And we even have a rough idea of what percentage of test takers will score in each category.
Applying the bell curve to the MAT
The median score on the MAT is 400. This represents the 50th percentile on the exam. It is also the “fat” part of our graph. That middle line would represent a score of 400 even. The standard deviation on the MAT is 25 points. Looking at our graph, we see that about 68 percent of test takers will score within 1 SD of 400; that’s a range of 375-425. 95 percent of all test takers will score within 2 SD’s of 400, between 350-450.
It’s worth noticing how quickly an increased scaled score improves your percentile rank. A score of 450 puts you in the 95th percentile. Yet, that is still 150 points from a max score of 600! Once we move to 3 SD’s above or below the median, we have accounted for 99.7 percent of all test takers. If you were to score a 475 on the MAT–still 125 points away from a max score–you would be in the top 0.2 percent of all test takers.
For comparison, male height also displays a bell curve. An MAT score 3 SD’s from the mean (475) is as rare as growing to be 6’10. I challenge you to think of men this tall outside of the NBA.