Market Research Tutorial: How to Determine Market Research Samples Sizes
At the outset, let's state that the terminology of market research samples sizes is a confusing one. It's often used interchangeably to describe two different sets of data used in a study.
Market research samples sizes often refer to both the total number of discrete records you will randomly draw from the population you are looking to understand. It also frequently refers to the total number of completes you need/want for your survey.
So, your first decision is how many completed surveys you will need.
This decision is driven by the level of precision (that is, the margin of error) you are willing to accept for each discrete portion of the results. Let's take an example, let's say you are surveying your customers and you know that 20% of your customers are "heavy" users, 30% are "medium" users and 50% are light users. If you want to be able to speak with statistical reliability about your heavy users as an individual (discrete) group (by taking just their results and analyzing them), you would need to determine how many surveys you need to completeamong this group. Using the table below, you see that if you have 100 "heavy" users, your margin of error is +/- 10%. You may be willing to accept this level of error, or you might not. It depends on how you will use the results. If your firm was looking at and testing a major change that will affect heavy users and is depending on the research to reveal their reaction, a 10% margin of error may be too much. For other uses, a 10% margin of error may be perfectly acceptable.
For this example, let's assume that a 10% margin of error on just the heavy users is okay. You would need to have a total of 100 completes among heavy users and then 150 medium users and 250 light users for a grand total of 500 completes if you take a simple random sample from your customer database.
Then, you will need approximately 15 customers for each complete you need, so you would need to pull a random sample of 7500 customers from your database.
Below is a table that displays various common market research samples sizes (referring to the number of completed surveys needed) andtheir associated margin of error with a 95% level of confidence. You will observe, you sharpie you!, that as the samplesize increases, the margin of error decreases. We're sure you noticed as well that in the study example above, a total of 500 completes yields a margin of error of just +/- 4.5%. So when you're talking about your entire customer base from the study above, you have a much lower margin of error than when talking about just heavy users.
Almost all market research studies are conducted based on a 95% level of confidence. This is the industry standard. Youcan use lower levels of confidence, again, depending on the criticality of the data and its intended use. However, mostorganizations want as much certainty as they can get and a 95% level of confidence is a decent balance between relative certainty and cost-effectiveness. (You can increase the level of confidence, but the number of completed surveys required skyrockets and so does the cost of the data collection.)
After you've determined your sample size (both of them!), you're ready to write your questionnaire. Onward!
Forward to Writing an Effective Questionnaire
Back to Sampling
Top of Market Research Samples Sizes Market Research Tutorial