A representative sample is one that accurately reflects the population being sampled. Acquiring a representative sample requires the use of random sampling techniques that avoid bias in the sample. Bias occurs when you are more likely to sample members of the population with certain characteristics than other other members of the population, causing one group to be over represented. The method you use will depend on the population size.

Create a numbered list of all members of the population and draw numbers out of a hat, or use a random number generating program, to get your sample. This method is called simple random sampling and works if you are sampling from a relatively small population, such as all the students at a school or employees at an office.

Use stratified sampling to preserve known groups that are important to your study. For example, if surveying employees in a company, you could sample from each department separately. Divide your sample proportionally between the department. For example, if 20 percent of employees work in sales, 20 percent in research and development and 60 percent in manufacturing, at total sample of 1,000 employees would include 200 from sales, 200 from research and development and 600 from manufacturing. Use simple random sampling in each subgroup.

Use cluster sampling to break down a large population into groups you can use simple random sampling on. For example, suppose that you want a random survey of all people in New York State. Start with a list of all counties in the state and use simple random sampling to select two counties. Then use simple random sampling to select two townships in each county. At that level you can use the phone book for each township as your numbered list for a final simple random sample. One quarter of the sample would come from each phone book.