Maybe visually would help. Definitions usually have a lot of mathematical jargon.
In: Mathematics
100 kids take a 100-question test. One kid only gets one question right, one gets two right, and so on up to 100. The kid who scored 50 scored better than 50% of the kids (because half of them scored less). The kid who scored 99 did better than 99% of the kids. You can flip that around, too. Let’s say you want to tailor classes to the kids, so you have a class with extra help for the kids that are struggling, a “normal” class, and an advanced class for the kids that are doing really well. The tutoring class and advanced class each have room for one fifth of the kids. What score do you need to be in the top 20%? You need at least an 80 to get into the advanced class. What’s the highest score you can have and still be in the bottom 20%? If you score under 20, you would be in the tutoring class.
It can be abstracted to less round numbers, less evenly distributed “scores” and a continuous variable instead of discrete test scores. Let’s say you measure the weight of 1,023 kids, to figure out who might need food assistance and who might need classes about healthy eating. Most of the kids weigh between 80-120 pounds. Maybe the area has a lot of poverty, so there are more kids that weigh less than that, say 65-80, than in the same 15-pound weight range from 120-135.
Since the weights aren’t evenly distributed or discrete like the test scores were, its harder to eyeball and say *this* is the bottom fifth and *that* is the top fifth. So you can use quantiles to calculate what the bottom and top 20% are. In this case the 20% quantile might be 70 pounds and the 80% quantile might be 120. There are the same number of kids in each group, but they have different weight ranges because the weights aren’t evenly distributed around the average of 100 pounds.
Latest Answers