So we want to collect some measurements. First, we need to understand some primary scales of measurement. This is useful when you're starting to think about designing a survey or a questionnaire, trying to gather data about the attitudes of a particular product or service. So let's dive into the primary scales. Essentially, there are four primary scales that you should all be aware of. There's nominal, ordinal, interval, and ratio scale data. They're all a little bit different. The first two are considered categorical variables, and then the last two are numerical scales. I shouldn't have said variables, but they are categorical scales and numerical scales. So let's look at them in turn. First, the nominal scale is probably one of the most crudest and simplest of measurement scales. It's a figurative labeling scheme in which the numbers will only serve as some label or a tag for classifying an object. So one common example would be to assign a number to a person's gender. So you could have zero for male, one for female, or vice versa. It really doesn't matter. But it's just something to distinguish between the two groups. Sometimes a nominal scale is considered a unique identifier for a person or an object. In that case, there's a strict one-to-one correspondence between the number and that object, so this is object one, this is object two, object three. Some common examples of a unique identifier might be your student identification number. In the United States, it would be your social security number, something like that. Next, we have what's called an ordinal scale. So an ordinal scale is ranking mechanism of categorical variables that's used to assign rank to the objects based on some characteristic. It does allow you to determine whether an object has more or less of some characteristic than another, but it doesn't tell you by how much. An example of an ordinal scale would be something like class rank. So are you a freshman in college, a sophomore, junior, or senior. So we could code that with numbers; 0, 1, 2, 3, 4, but it doesn't really tell you how far apart they are. We don't have a concrete understanding of the underlying rules that determine your class status. So is the amount of the distance between a freshman and sophomore the same as between a junior and senior? We don't really have a good sense of that. But order does matter. So we know that sophomores come after freshmen, juniors come after sophomores, and seniors come after juniors, so that's why order does matter. You might want to pause the video for a second and see if you can come up with other examples. I'll pause for a second and let you think about that. So another example might be military rank, private, corporal, sergeant. Another example might be your classification in an organization, a rookie employee, supervisor, manager, senior manager, that kind of thing. So that wraps up the categorical variables. There's the nominal and ordinal. Really, the only difference is that nominal scale data, order does not matter; male, female, for zero and one, that's common example, eye color, red or blue, red, who has red eyes? Black, or brown, or blue. So those are three nominal colors of eyes that you might see. It really doesn't matter if blue comes before black or brown comes after blue, etc. All I know is also categorical variables where the order does matter; class rank, military rank, things like that. Now we have interval scale data. This is where numbers do become important. An interval scale data ranks the attributes such that the distances are equally distanced on the scale, and that's how the characteristic is measured. One thing to note is that all the information that's contained in an ordinal scale is also contained in an interval scale, but the key difference is that with an interval scale, you can compare the differences between two points, between objects. However, the location of the zero point or the zero reference point is not important, it's not fixed. So what's an example of something that's considered interval scale data? Temperature is an interval scale data. So I'm going to talk in Fahrenheit for a minute. If something is 32 degrees Fahrenheit, and then a couple hours later, let's say it's the temperature, it's 34 degrees Fahrenheit, we can say that there was a difference of two degrees Fahrenheit. That's two degrees Fahrenheit is the same difference as between 60 and 62 degrees Fahrenheit. So we can take the differences and compare those differences, and comparing those differences would have meaning. But if you think about the zero point, it's arbitrary. What is zero degrees Fahrenheit? It's an arbitrary point on the scale. It really does not have meaning. Whereas in the Celsius scale, the zero point, it's tied to the freezing point of water, but it's also an arbitrary point. The reason it's called interval scale data is that you can only look at intervals, the difference between two data points. What you cannot do is use ratios. For example, yesterday was 10 degrees outside, today it's 20 degrees outside, and you would not, in normal conversation or ever, say today is twice as cold as it was yesterday. You can say it was 10 degrees warmer, but you would not say it's twice as warm or twice as cold. So you wouldn't say these types of things. The zero point is arbitrary and it doesn't really have meaning. If the zero point does have meaning, you have something that's called ratio scale data. So ratio scale data possesses all the characteristics of nominal data, ordinal data, and interval scale data. So it has order. You can compare the difference between two points on a interval scale and ratio scale, but zero does have meaning. There is a key element to the zero scale. So with ratio scale, we can rank the objects, we can look at the differences. Some common examples would include something like weight. It really doesn't matter what unit of measurement you use. So if something is one kilogram and something else is two kilograms, you would say the second object weighing two kilograms weighs twice as heavy as the first object. So you can do these types of comparisons. Zero does have meaning. In that sense, it does have a reference point. Age is another reference point. So if someone is five years old and someone's 10 years old, you can say that the 10-year-old is twice as old as the five-year-old. So when you think about doing these ratios of a over b, then you know that you have something that's in ratio scale format. So why is this important? We've talked about nominal data, ordinal data, ratio scale data, and interval data. Those are the four types of data. The first two; nominal and ordinal are categorical, and the next two; interval and ratio scale are numerical. The ideal form of data that you want because you have more things you can do with that data is ratio scale data, that's the best, and the one that has the least amount of granularity is nominal data, which just has categories that don't even have order. One thing to note is that you can always go from ratio scale data and work your way to less and less and less refinement. Meaning, you can go from ratio scale data to interval scale data to ordinal data to nominal data if you so desire. So there might be, for example, you might have a continuous ratio scale set of information, and then you might arbitrarily set break points. Say, if it's under a 100, that's considered okay. Then, if it's over a 100, that's considered great. So now you have these two ordinal categories, okay and great. Obviously, you can break it down some more. Then, you could even make it just okay and great and not have any more detail around that. The key is that if you're given a piece of data that's in ordinal form; good, bad. Let's do it in order; fair, good, great. So those are three categories that you might want to assess attitudes about a product or service. You really can't get any more granular understanding of that data, and you can't go from that data point to ratio scale data, but you can go from ratio scale data down to ordinal or even nominal classes. So that wraps up the primary scales.