In the previous video you learned about how you can use descriptive statistics to describe a data sample. In this video will learn about the other main category of statistics, inferential statistics. Inferential statistics uses sample data to make inferences about the population. It's a way of drawing conclusions from our sample data and then generalizing these conclusions to a larger population. Since we can't usually study an entire population, we collect a sample that's representative of that entire population, and then calculate measures about that sample. We then use the sample characteristics to make generalizations or inferences about the population. In order to make accurate inferences, the sample has to be representative of the actual population. A representative sample is one where each and every member of that population has an equal and mutually exclusive chance of being selected as part of our sample. This means that we're not biased and how the sample is chosen from the population. We wouldn't want to choose only the adults that we know in our town for our hobby survey, because this wouldn't reflect the true nature of the whole population and their hobbies. Maybe everyone I know in Oakland, California is really into kayaking, but that doesn't mean that I can say that's true for everyone in the entire United States. There are many sampling methods that are used to ensure that samples are representative and unbiased. I'm not going to go into the details of sampling methods here, it's just important to remember that selecting a good sample is critical for making inferences about a population. Despite our best efforts at making sure samples are representative of the population, samples would never be expected to perfectly represent the population. This is where inferential statistics comes in. We can use them to better understand these errors. In turn, this allows us to determine whether the patterns we observe in our sample data actually generalized to the population, and whether these sample statistics adequately support our hypothesis about population. Now that you've learned about descriptive and inferential statistics, you might be wondering when you would use one or the other. A good way to decide is to ask yourself whether you're trying to just describe a sample, or if you're trying to prove or disprove a hypothesis, or make some sort of prediction about the actual population. To go back to our fish example, let's say you've observed and recorded measurements of 100 fish in Bass Lake every month for the past year. You want to describe these measurements and maybe create a visualization of the mean and median sizes of the fish in your sample overtime. This would fall into the category of descriptive statistics because you're just describing the sample. But you're probably not just randomly collecting data on some fish in the Lake, just for fun. You actually would want to know if the entire population of fish in the Lake has changed significantly in size over the past year. Maybe you even want to predict for each month of the next year what the mean and median sizes of the fish might be. This is when you would use inferential statistics to extrapolate or generalize from what you observe in your sample to the whole population.