If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

# Threshold for low percentile

Find the cutoff for a given lower percentile in a normal distribution.

## Want to join the conversation?

• At , Sal mentioned that he wanted to find a z-score that represents a percentile that is barely less than 10%. What if the problem asked for top ten percent? Would we be looking for barely under or over 90%?
(10 votes)
• for the top 10 percentile, you will look for barely over 90%,because suppose if you choose 89.98% then that is no longer in the top 10 %, however, if you choose 90.01% it is within top 10%.
(22 votes)
• Why does Amelia vaule wait time over quality of food? That is the one question that remains after the video...
(14 votes)
• How can you get the actual estimate of % on the density curve? I don't get how 10% was marked on that specific location (same goes with the previous lesson's 30%). Any guides that led to this?
(5 votes)
• Good question Josh, Don't worry about this! As long as you know that you're looking for the 10% below (0.10) you should be able to find the z-score in the chart.
(5 votes)
• Hi, I didn't understand one thing; since the distribution is normal (symetric) why you didn't take the second part (side) where the wait time is <10% I mean 185+1.29*11=199.19.
(1 vote)
• Maybe this is my programmer-brain, but I actually misread the question too at first, because I reasoned that the "bottom 10%" was going to be the restaurants with the longest waiting times. When I realized that it would be weird to go to restaurants that had the longest waiting times, I assumed that the bottom 10% meaned the bottom 10% of the distribution.

To be fair, the question doesn't specify!
(7 votes)
• When you put the video speed on `0.25`, it sounds like everything Sal says is a question. XD
(2 votes)
• Why is the z score -1.29?
(1 vote)
• What videos should i search for if i want to know the calculator version for this kind of
problem?
(1 vote)
• The type of calculator you have and then the function normalcdf
(1 vote)
• At . how did Sal know where the 10th percentile was, in relation to the SD?
(1 vote)
• You can calculate the probability from the cdf of a normal distribution. This is given in a table. From there you see the 10th percentile is between -2*(sigma) and -1*(sigma).
(1 vote)
• Ah!I found a problem! Isn't Z scores is from 0 to Z? so for this question 10% is actually 0.5-0.1=0.4 in z scores. We need to check the table for 0.4 rather than 0.1
(1 vote)

## Video transcript

- [Instructor] The distribution of average wait times in drive-through restaurant lines in one town was approximately normal with mean of 185 seconds and standard deviation of 11 seconds. Amelia only likes to use the drive-through for restaurants where the average wait time is in the bottom 10% for that town. What is the maximum average wait time for restaurants where Amelia likes to use the drive-through? Round to the nearest whole second. Like always, if you feel like you can tackle this, pause this video and try to do so. I'm assuming you've paused it. Now let's work through this together. So let's think about what's going on. They're telling us that the distribution of average wait times is approximately normal, so let's get a visualization of a normal distribution. And they tell us several things about this normal distribution. They tell us that the mean is 185 seconds, so that's 185 there, the standard deviation is 11 seconds. So, for example, this is gonna be 11 more than the mean, so this would be 196 seconds, this would be another 11. Each of these dotted lines are one standard deviation more. So this would be 207. This would be 11 seconds less than the mean, so this would be 174, and so on and so forth. And we want to find the maximum average wait time for restaurants where Amelia likes to use the drive-through. Well, what are those restaurants? That's where the average wait time is in the bottom 10% for that town. So how do we think about it? Well, there's going to be some value, let me mark it off right over here in this red color, so we're going to have some threshold value right over here where this is anything that level or lower is going to be in the bottom 10%. Or another way to think about it is, this is the largest wait time for which you are still in the bottom 10%. And so this area right over here is going to be 10% of the total, or it's going to be 0.10. So the way we can tackle this is we can get up a z-table and figure out what z-score gives us a proportion of only 0.10 being less than that z-score, and then using that z-score, we can figure out this value, the actual wait time. So let's get our z-table out, and since we know that this is below the mean, the mean would be the 50th percentile, we know we're gonna have a negative z-score. So I'm gonna take out the part of the table that has the negative z-scores on it. And remember, we're looking for 10%, but we don't want to go beyond 10%. We want to be sure that that value is within the 10th percentile, that any higher will be out of the 10th percentile. So let's see, when we have theses really negative z's, so far it only gets, it only doesn't even get to the first percentile yet. So let's scroll down a little bit and let's remember as we do so that this is zero in the hundreds place, one, two, three, four, five, six, seven, eight, nine, so let's remember those columns. So let's see, if we are at a z-score of negative 1.28, remember this is the hundreds is zero, one, two, three, four, five, six, seven, eight, so this right over here is a z-score of negative 1.28 and that's a little bit crossing the 10th percentile. But if we get a little bit more negative than that, we are in the 10th percentile. So this is negative 1.29, and this does seem to be the highest z-score for which we are within the 10th percentile. So negative 1.29 is our z-score. So this is z equals negative 1.29, and if we want to figure out the actual value for that, we would start with the mean, which is 185, and then we would say, "Well, we want to go 1.29 standard deviations "below the mean." The negative says we're going below the mean. So we could say minus 1.29 times the standard deviation and they tell us up here the standard deviation is 11 seconds. So it's going to be 1.29 times 11. And this is going to be equal to 1.29 times 11 is equal to 14.19. And then I'll make that negative, and then add that to 185. Plus 185 is equal to 170.81. 170.81. Now they say round to the nearest whole second. There's a couple of ways to think about it. If you really want to ensure that you're not gonna cross that 10th percentile, you might want to round to the nearest second that is below this threshold. So you might say that this is approximately 170 seconds. If you were to just round normally this would go to 171, but just by doing that, you might have crossed the threshold. But in all likelihood for this application where someone is concerned about wait time at drive-through restaurants, that difference in a second between 170 and 171 is not going to be mission critical, so to speak. But you could safely say that 170 or 171 seconds will meet Amelia's needs.