Research the population of the study. If the population that you wish to examine is small, then you may be able to complete a census of everyone within your target group. If the population to be studied is much larger, then sampling the population is more efficient.
Decide on the method and design of the study. The design and method of study needs to fit within your budget. A couple of questions that you should answer before beginning the analysis include: How are you going to extract the data from the sample population? If you are going to survey the participants with a questionnaire, how is that questionnaire going to be distributed? For example, placing a survey on the Internet is much less expensive than individually mailing or polling all of the individuals.
Determine the level of precision of the study. Generally speaking, the more individuals who participate, the more accurate the results. The level of precision is how close the sample data is to the total population. The difference between the sample size and total population is known as sampling error. This metric is usually expressed in terms of a percentage. For example, if study shows that 70 percent of a population uses a product and the precision rate is plus or minus 5 percent, then it is safe to conclude that between 65 and 75 percent of the population uses that product. Estimate a level of precision.
Identify the confidence level. The confidence level is the measurement of how likely your sample size equals the true population. For example, a confidence level of 95 percent would mean that if any 100 members of the total population were surveyed, 95 percent would be within the range of precision identified earlier. A good estimate for confidence level is between 90 and 98 percent.
Estimate variability. The variability of a study measures the distribution of attributes in the given population. In other words, if the population is heterogeneous, then a larger sample size is required to fall within the level of precision. If a population is very similar, then a smaller sample size may work. Estimate the the size of the smaller characteristic you wish to measure to determine variability. For example, if you estimate that 15 percent of the population uses a specific product and 85 percent does not, then the variability would be 0.15 or 15 percent.
Estimate your response rate. To estimate your response rate, consider the method of your study and the population. The easier it is for people to respond to your survey, the higher your response rate will be. Additionally, more contacts with the various populations will also increase your response rate. The base sample size is the minimum amount of responses needed for your study. Because not everyone will participate, the sample size may be increased to cover anyone who elects not to participate.
Calculate your representative sample size. Because there are numerous formulas to calculate sample size, and the formulas may differ depending on the type of survey, check out online sample size calculators to determine your sample size. Each of the estimations above are used in the most of the calculators.