Question 157740
Say you want to take a poll of people and see how much money they make. 
Let's say you ask 1000 people. 
Now you need to decide how to represent this data. 
You can take every value and plot it on a graph. 
Not very helpful.
You can bracket the results and make a frequency distribution.
If you have a random sample, you'll most likely have a lot in the middle (say $50,000/year), fewer on the low and high edges. 
You don't want to throw that data out but what's the best way to characterize it.
Say you had 5 high school and college students in the poll and their yearly pay was $1,200, $2,500, $3,200, $4,300, and $8,100. 
You don't want to have a bracket from $0-$1000, $1000-$2000, and so on, because you don't gain much information. 
So you make a bracket $10,000 and under, to capture those 5 data points.That's not really open ended because the bracket starts at $0.
Similarly on the high side.
Say you polled a doctor, a lawyer, Bill Gates and they made $220,000, $450,000, and $37 billion. 
You may want to have an open ended upper distribution where they are 3 data points in the $200,000 and over bracket. 

To make a long story short, they're important because it gives you a way to quickly and effectively represent data.