SAT Statistics: Mean, Median and Spread

Mean, median, mode, or range can each describe the same set of numbers, but every one of them answers something different. Mean adds everything up and divides by how many values there are. Median is the middle value once everything's in order. Range is just the gap between the highest and the lowest. Mode is the value that shows up most often. On the SAT, Problem-Solving and Data Analysis questions hand you a list, a frequency table, a dot plot, or a histogram, then ask for one of these, or ask how one changes when the data does.

Questions also ask you to compare spread between two data sets, usually using the phrase "standard deviation." Don't panic. Digital SAT questions almost never ask you to calculate it by hand. They ask which set is more spread out, meaning which set's values sit farther from their own average. That's a visual, conceptual call, not a formula to memorize.

Here's a small data set: five practice-quiz scores, 62, 65, 68, 70, 72. Add them up and divide by five: a mean of 67.4. Sort them and pick the middle one: a median of 68. Subtract the smallest from the largest and the range comes out to 10. Nothing dramatic there. Scores cluster close together.

Now add a sixth score: 95, way above the rest. Suddenly the mean jumps to 72, a full 4.6 points higher. Barely anything happens to the median: it climbs from 68 to just 69, since it's based on position, not size. Range is a different story: it explodes from 10 to 33. One extreme number can wreck the mean and the range while leaving the median almost untouched. That gap is exactly what the SAT tests when it asks which measure changes most after a value gets added or dropped.

Frequency tables trip people up because they hide the actual list of values behind a count. Say ten students each timed how long they spent on one tricky question: three took 1 minute, five took 2 minutes, two took 3 minutes. Written as a table, that's three rows. Written as the list it represents, it's ten numbers: 1, 1, 1, 2, 2, 2, 2, 2, 3, 3.

Once it's a list, the mean is just the total divided by 10: three times 1, plus five times 2, plus two times 3, all over 10. That comes out to 1.9 minutes. Sort the values and the median is the average of the fifth and sixth ones, both of which happen to be 2, so the median is 2. Skip the expansion step and it's easy to average the frequencies instead of the values, or to grab the middle row of the table instead of the middle data point. Expand first, then compute.

Picture two sets of scores with the same mean. Set A: 48, 49, 50, 51, 52. Set B: 30, 40, 50, 60, 70. Both average out to exactly 50. But Set A's values huddle close to that average, while Set B's values swing far on either side. Set B has the larger standard deviation, even though neither set is bigger than the other in any obvious way.

That's the trap built into these questions. Plenty of students assume the set with bigger numbers has the bigger spread, or that spread has something to do with the size of the average itself. It doesn't. Standard deviation only cares about distance from the mean. On a dot plot, this shows up as how tightly the dots cluster versus how far they fan out. If a question shows two plots or two tables and asks which one is more variable, picture the dots. Which group hugs the center? Which one spreads wide? That's the answer.

Work with the total, not the average, whenever a question adds or removes a value. If the mean of 5 numbers is 20, the total is 100. Add a sixth number and the new total tells you the new mean fast, without rebuilding the whole set from scratch. That trick alone saves time on half the "what happens to the mean" questions.

For median questions, reorder the list first. Always. Sometimes the SAT lists values out of order on purpose, hoping someone grabs the middle position in the table instead of the real middle value once everything's sorted. Write the numbers out from smallest to largest, then the median takes two seconds to find. On these questions, speed comes from doing the one extra step, not skipping it.