How you choose a sample is a big factor in whether the sample is a good representation of the population. They're several methods that can help you but today we'll go over the most common methods for selecting a sample and when to use them. Now, there are two general categories of methods: Probability sampling and Non-probability sampling. Probability sampling is randomized, so you are more likely to get results that represent a whole population. Non-probability sampling does not randomize and is more likely to capture several members of one group of the population, while missing other groups entirely. These often lead to poor representations of the population, inaccurate analysis, and unhelpful insights in your data. Imagine if you only collected surveys in a store during the workweek from 9:00 AM to 10:00 AM. You won't get any data from customers who worked during the day coming in the evening, or those who sleep in or stop in in the afternoon. This sample would not accurately represent the population of customers as a whole, and that is why in this course we will only be teaching probability sampling methods. In this video, we will cover four types of sampling. Simple random sampling, systematic sampling, stratified sampling, and cluster sampling. Simple random sampling is the most direct method so we'll start there. In simple random sampling, everyone in a population has an equal chance of being selected. You can think of this as putting everyone's name in a hat and drawing five names out. Well, this is the most random sampling method, it can only be used with small populations. For example, if a gym wants to survey current members, they already have all of the information for the entire population so selecting a few names at random would be a good representation of their gym members. The next approach is systematic sampling, which is used when your population is brought to you in a series or one-by-one. You survey the first individual, then the next individual, and so on or you can break the series into a regular interval, like choosing every fifth person to survey or every 10th. For example, you want to survey people who go to the gym on Saturdays to see if they would be interested in a special event that only takes place during the week. To do this, you wait by the front door of the gym and survey every fourth person who enters. It doesn't matter what number you choose for the interval but it is important that you stick to it so that you can achieve a randomized sampling. The next approach is stratified sampling, which ensures that the sample has the same ratio of a specific variable as the population. You should use this method when you know the demographic ratio of your population and you want to make sure every group is represented. For example, the gym sells protein powder that helps customers bulk up. There's a new powder flavor that they want to gauge the interest in. The gym knows that 90 percent of their customers that buy the powder are male, but they want to be careful not to alienate their few female members. If you were to do a simple random sample, you may not get any female members because there's only one for every nine male customers. Stratified sampling means that in order to get a representative sample of 100, you would divide your population. In this case, protein powder purchasers by gender and then randomly sample 90 customers from the male group, and randomly sample 10 customers from the female group. This will give you a sample with the exact same proportions as your population. Our last approach is cluster sampling. You would use cluster sampling when your population is broken down into subgroups. After groups are selected, you can then choose individuals within each group with a second sampling method if you like. For example, the gym wants to survey customers who take exercise classes, to see if they'd be interested in special deals. They're many different classes and you can think of each class as a subgroup. With cluster sampling, you would pick classes at random and then survey them. While strict cluster analysis sampling would survey everyone in the selected classes, any of the other methods can be used in conjunction with cluster sampling as well. For instance, after selecting exercise classes, you can randomly select customers from within that class with simple random sampling or you can select every third person who enters that class with systematic sampling, or you can make sure your sample for every class has the same gender ratio as your population with stratified sampling. For cluster sampling, you just need to remember to break your population down into groups and then choose a few groups at random. Now that you know the four main sampling techniques, let's look at one more example. DCB Cleaning is a company that manages cleaning and janitorial services. The majority of their customers are small and medium businesses, but they do have a few customers who are large companies. DCB Cleaning is considering a new service and wants to survey a few of their customers to see if there would be any interest. They do want to make sure that they survey companies of every size, including large companies, even though they do not serve as many. What is the best sampling method for this situation? You would use stratified sampling because DCB Cleaning wanted to make sure they got information from every group in their population, including large companies which are a minority. They're many different ways to apply these methods. Get creative. Think of a survey you might want to give, then consider which method might be the best way to gather your sample.