Advertisement
FTCE General Knowledge Mathematics
This Core Video focuses on An Overview of Statistics. It is part of our Free FTCE General Knowledge Mathematics Prep Course, which also includes realistic, full-length practice tests with video explanations for every multiple choice question.
Done watching this? Return to the course homepage to access your other free videos.
Return to Course HomepageAdvertisement
Video Transcript
Below is a full text transcript of our FTCE General Knowledge Mathematics video above.
- So let's talk a little bit about, just a general overview of statistics. A statistical question is typically one that anticipates variability in the data or answers. So if you had asked what age is my dad? That is not a statistical question because there's only one, one piece of datum you're gonna get in response from that. But if you ask what is the average age of people in my town, that is a statistical question. You're going to get a large variation in the answers and then you're gonna find the mean of those answers.
So that's sort of how we define a statistical question. The data collected to answer a statistical question can be described based on its center, spread, and overall shape. What we're talking about the center, the measures of central tendency that are most often used are mean or median. When we're talking about spread, we're talking about how closely grouped is the data around the median or mean, and are there clusters elsewhere as well? Also, is that spread, when we're talking about the overall shape, is it sort of regular?
Like are all the data values spread out evenly? Or are the clustered pretty tightly around the mean and there's just a few little outliers out beyond that. So we can look at the overall shape, the spread, and the center, and if the spread is sort of wide, that tells us that the measure of central tendency is not going to be necessarily as representative of the overall population, 'cause there's a lot of sort of variability there. But if everything's closely grouped around the mean or median, then the mean or median is more representative of the overall population.
So these are just some ways that we can kind of help understand and interpret the statistical values that we get. So now let's talk a little bit about how we actually design surveys. A survey, for starters, is just a sampling of facts and figures and opinions used to represent a larger population. So, for example, you wanna know how all of California's voters might vote on a given issue, you might survey 5,000 representative people and try to extrapolate from that how all of California's voters are gonna vote based on how that sampling of people said they were gonna vote.
So, another example, we might survey people at a shopping mall to determine what percentage of California's population works in or plans to work in the field of education. Let's imagine we were doing that particular survey. If this was gonna be a good survey, the data should be random and representative. We need to understand what both of those mean because sometimes they're gonna ask you about sources of bias, and when you're trying to identify a source of bias, typically the bias comes from the fact that it's either not random, not representative, or both.
Random, meaning that it should be not focused on one group that wouldn't encapsulate the entire population you're trying to represent. We'll see an example in a little bit. And it needs to be representative, meaning there's enough people that it would represent that population. You survey only a couple people, even if those are random people, it's really not enough to represent a very large population. So surveys need to be random and representative. Bias occurs when surveys are not random or representative.
So, one source of bias is that a sample is too limited. Let's imagine that our mall surveyor goes out and he surveys two people at the mall. He asks, "Do you work in education?" No. "Do you work in education?" No. And so he says, okay, then I guess 0% of California's population works in the field of education. So this doesn't, this doesn't make sense. So what's the problem? He didn't survey enough people. His sample is too limited and that's created bias here. It's not really a representative sample. Another source of bias is that the sample is skewed.
So let's say our mall surveyor goes and he surveys 2,000 people, but he does it on weekdays between 9:00 a.m. and 3:00 p.m. and he says, "Do you work in education?" No, no, no, no. Everybody says no. So again, he says, I guess 0% of the state population works in education. Again, this is clearly not true. He had a larger sample, it wasn't too limited this time. It was a fairly broad sample, but the problem is the sample wasn't really random. It was skewed because people who work in education typically are teaching school during that time period.
They wouldn't be at the mall on a weekday between 9:00 a.m. and 3:00 p.m. in order to take the survey, so the people who were there were not really the people that he was trying to sample. They didn't really represent that population. Sometimes you could have a sample that is not representative and not random. Could be too limited and could be skewed. He could survey two people and he could do that between 9:00 a.m. and 3:00 p.m. on weekdays, so then you have both of the problems. So, when you're looking for the sources of bias in a question, look for one or both of those two that we just talked about.
So when examining the process underlying statistical experiments, several useful questions to ask include how many observations took place? Usually the more observations you have, the more reliable the data. What is the attribute being investigated? How is it measured? So, the ways that we're measuring something can affect the kind of data that we get, 'cause, you know, certain units are going to give us sort of different levels of accuracy, for example, if we were measuring people's heights in centimeters versus in millimeters, we would have a different level of accuracy.
So, what is the attribute? What way are we measuring it? What was the context in which the data was gathered? If we're measuring the height of people in a town to figure out what the average height is, were we measuring people with their shoes on? Were we measuring people at certain times of day and in certain environments where the might stand up taller, slump more? Et cetera. If the data is bivariate, meaning that there are multiple variables that we are accounting for, is there any correlation between the variables.
Do we find that taller people tend to have brown hair? So if we're looking at hair color and height, is there a correlation? So those kinds of things might tell us something about a population as well. There's several ways in which we can describe the distribution of data. The first one is by quantifying the center, and typically we do that via the mean or the median. With the mean we add up all the members of the set and we divide by the number of members that are in that set. With the median, we put them in order from smallest to largest and we take the center value, or if it's an even number, we split the difference between the two centermost values.
So those are two primary ways that we describe the center of a distribution. Another thing that we look at when we're talking about distribution is the variability. So, how much sort of variation is there in the data relative to that mean or median? And two ways that you should be aware of for the test are interquartile range or mean absolute deviation. So let's talk a little bit about each of those. So first of all, interquartile range. With interquartile range, quartile, quarters. So we have a set of data.
First quarter, second quarter, third quarter, fourth quarter. Imagine that those are all basically drawn equally. So that's first quarter, second quarter, third quarter, fourth quarter. So, the median would be here in the middle and the interquartile range basically looks at the two quarters that are closest to the median. So what we're really doing is subtracting this value, the sort of Q1 value, from this value, the Q3 value, to get this overall value. Now that is going to tell us, if that number is fairly large, that there's a pretty wide distribution from the median.
If that number is pretty small, that's gonna tell us that there's a pretty tight distribution around the median. And we've essentially eliminated a lot of the outliers that existed in Q1 and Q4 by doing this. So let's look at an actual example of how you might encounter this on the test. Let's say that they give us a set of numbers that are a response to some kind of survey. So whatever it happens to be, the number of gumballs that the average person likes to eat. So they give us a bunch of data. Three, five gumballs, somebody says seven gumballs, somebody else says nine, someone says seven, somebody says 2, someone else says nine.
So we've asked one, two, three, four, five, six, seven people how many gumballs they like to eat per day. So, this is our data. Now we're gonna put it into order. So we're gonna go two, three, five, seven, seven, nine, nine. So we've now ordered that data, and we're gonna divide it into quarters. So if we draw a line here, that's the first quarter. That's the second quarter or median. This is the third quarter, so this would be the demarcation of the end of the third quarter. This would be the fourth quarter here.
So first quarter, second quarter, third quarter, fourth quarter. And these are demarcation lines at the end of the first quarter, the beginning of the fourth quarter or the end of the third quarter, however you wanna describe that. So, to figure out what interquartile is, remember it's these two quarters that are closest to the median, where you're just gonna subtract three from nine. It's as simple as that. So the interquartile range in this case is six. So the smaller this number is, the more tightly grouped that data would be around the median.
And the larger this number is, the less tightly grouped that data is around the median. So that's one way of describing variability. Another way is absolute mean, or mean absolute deviation. With mean absolute deviation, let's say we had the same set of numbers. And again, we just have them in order here. Now we wanna find the mean. So we're gonna add this up. Two plus three is five, plus five is 10, plus seven is 17, plus seven is 24, plus nine is 33, plus nine is 42. So we have 42 as our total and if we're trying to find the mean, we wanna divide that total by the number of members, there's seven members in this set of data.
42 divided by seven gives us six, so six is our mean. Now, we wanna see what the deviation, the mean deviation is from the mean, so we're gonna look at how each of these numbers deviates from the mean of six, and we're gonna find the mean of that deviation from the mean. So, what is the average deviation away from that measure of central tendency. And again, we're saying how closely grouped is this data around our center point. So the way we're gonna wanna do that is to sorta make a list of these numbers, and we can just leave them listed the way we want or you could make a list up and down, whatever you prefer.
But you want to sort of subtract each of these numbers from our mean and then we wanna see how far away that is from the mean. Up here, I'm gonna subtract the numbers from the mean, so two from six is gonna give us four. Three from six is gonna give us three. Five from six is gonna give us one. Seven from six is gonna give us negative one, but these are always gonna be positive values, and that's why we call it mean absolute deviation. Because we're taking the absolute value of the number. So that, remember the absolute value sign makes negative one become one.
So that's why we're talking about absolute value. So, six minus seven is negative one, but we're dropping the negative, we're just getting one. Six minus seven again, again one. So that's how far away seven is from six. Makes sense, right? And then, we subtract nine from six, negative three or just three, so the difference between six and nine is three. And again, the difference is three. So these values up here are the deviations from the mean for each of these members of our data set. Now we wanna find the average of these, so we just do the same thing.
We're finding the mean of the deviations. So four plus three is seven, plus one is eight, plus one is nine, plus one is 10, plus three is 13, plus three is 16, so now we're gonna divide these 16 by seven, because there's seven numbers and you'll need to do this on your onscreen calculator. So 16 divided by seven equals approximately, it's gonna be approximately 2.29. So 2.29 is our mean absolute deviation. Meaning it is the average of the deviation of each of the data set numbers from the mean. So what, how far on average is everything away from the center.
If this number's big, that means things are pretty far away from the center on average. If this number's small, it means that things are pretty tightly clustered around the center and therefore that that measure of central tendency is a good representative of the overall set, or a close approximation of the overall set. In other words, it speaks a little more closely for all the members of the set. So those are a few ways of talking about variability. How much variation is there in the data? We also might describe the overall pattern, so we might describe certain clusters that we get a cluster at this point and a cluster at this point and a cluster on the mean, or that we have a very even distribution or that we have a strong clustering at the outlier regions.
We also want to note any striking deviations from the overall pattern. So if we see that there's a pretty even distribution and then there's a space where there's just a huge gap, nobody likes this certain number of gumballs, then that's notable too in describing the overall distribution of the data.